model family
Gemma 4 logo

Gemma 4

Published by Google. 7 members measured on real machines, across 40 builds. Every number below was signed by the machine that produced it — none of it is from a model card.

what it takes, what it returns
image + text → texttext → text

Declared by the network’s catalog, not measured — a signature says what this family would be asked. Whether it answered is the signed capability flags on each member.

40 of 69 builds measured · 29 still open
what each build proved
12b · Q8_0
e2b · Q4_K_M · unsloth
e2b · Q4_K_M · unsloth
e2b · Q4_K_M · ollama
e2b · IQ4_XS
e2b · Q4_0
12b · Q4_K_M
26b-a4b · Q8_0
e2b · BF16
e2b · Q8_0
e4b · Q4_K_M · ollama
e2b~bartowski · BF16
e2b~bartowski · IQ2_M
e2b~bartowski · IQ3_XS
e2b~bartowski · Q3_K_L
e2b~bartowski · Q4_0
e2b~bartowski · Q4_K_M
e2b~bartowski · Q8_0
12b · Q4_0
12b · BF16
12b · IQ4_XS
31b · Q4_0
31b · Q4_K_M
e4b · BF16
e4b · IQ4_XS
e4b · Q4_0
e4b · Q4_K_M · unsloth
e4b · Q8_0
12b~yuxinlu1 · Q3_K_M
12b~yuxinlu1 · Q4_K_M
12b~yuxinlu1 · Q6_K
12b~yuxinlu1 · Q8_0
e2b · Q3_K_M
e2b · IQ4_NL
e2b · Q3_K_S
e4b · Q3_K_M
e2b · e2b-it-qat
e4b · e4b-it-qat
e2b
e2b · Q2_K_XL
12b · Q8_0 · Tool calling — proved12b · Q8_0 · Tool loop finished — proved12b · Q8_0 · Faithful to tool results — proved12b · Q8_0 · Chained tool calls — proved12b · Q8_0 · Structured output — proved12b · Q8_0 · Sustained output — proved12b · Q8_0 · Extended thinking — proved12b · Q8_0 · Vision — proved12b · Q8_0 · Counts in images — proved12b · Q8_0 · Multiple images at once — proved12b · Q8_0 · Spatial reasoning — proved12b · Q8_0 · Writes code — proved12b · Q8_0 · Reasons about code — proved12b · Q8_0 · Synthesises sources — proved12b · Q8_0 · Follows instructions — provede2b · Q4_K_M · unsloth · Tool calling — provede2b · Q4_K_M · unsloth · Tool loop finished — provede2b · Q4_K_M · unsloth · Faithful to tool results — provede2b · Q4_K_M · unsloth · Chained tool calls — provede2b · Q4_K_M · unsloth · Structured output — provede2b · Q4_K_M · unsloth · Sustained output — provede2b · Q4_K_M · unsloth · Extended thinking — provede2b · Q4_K_M · unsloth · Vision — provede2b · Q4_K_M · unsloth · Counts in images — provede2b · Q4_K_M · unsloth · Multiple images at once — provede2b · Q4_K_M · unsloth · Spatial reasoning — provede2b · Q4_K_M · unsloth · Writes code — provede2b · Q4_K_M · unsloth · Reasons about code — provede2b · Q4_K_M · unsloth · Synthesises sources — provede2b · Q4_K_M · unsloth · Follows instructions — provede2b · Q4_K_M · unsloth · Tool calling — provede2b · Q4_K_M · unsloth · Tool loop finished — provede2b · Q4_K_M · unsloth · Faithful to tool results — provede2b · Q4_K_M · unsloth · Chained tool calls — provede2b · Q4_K_M · unsloth · Structured output — provede2b · Q4_K_M · unsloth · Sustained output — provede2b · Q4_K_M · unsloth · Extended thinking — provede2b · Q4_K_M · unsloth · Vision — provede2b · Q4_K_M · unsloth · Counts in images — provede2b · Q4_K_M · unsloth · Multiple images at once — provede2b · Q4_K_M · unsloth · Spatial reasoning — provede2b · Q4_K_M · unsloth · Writes code — provede2b · Q4_K_M · unsloth · Reasons about code — provede2b · Q4_K_M · unsloth · Synthesises sources — provede2b · Q4_K_M · unsloth · Follows instructions — provede2b · Q4_K_M · ollama · Tool calling — provede2b · Q4_K_M · ollama · Tool loop finished — provede2b · Q4_K_M · ollama · Faithful to tool results — provede2b · Q4_K_M · ollama · Chained tool calls — provede2b · Q4_K_M · ollama · Structured output — provede2b · Q4_K_M · ollama · Sustained output — provede2b · Q4_K_M · ollama · Extended thinking — provede2b · Q4_K_M · ollama · Vision — provede2b · Q4_K_M · ollama · Counts in images — provede2b · Q4_K_M · ollama · Multiple images at once — provede2b · Q4_K_M · ollama · Spatial reasoning — provede2b · Q4_K_M · ollama · Writes code — provede2b · Q4_K_M · ollama · Reasons about code — provede2b · Q4_K_M · ollama · Synthesises sources — provede2b · Q4_K_M · ollama · Follows instructions — provede2b · IQ4_XS · Tool calling — provede2b · IQ4_XS · Tool loop finished — provede2b · IQ4_XS · Faithful to tool results — provede2b · IQ4_XS · Chained tool calls — provede2b · IQ4_XS · Structured output — provede2b · IQ4_XS · Sustained output — provede2b · IQ4_XS · Extended thinking — provede2b · IQ4_XS · Vision — provede2b · IQ4_XS · Counts in images — provede2b · IQ4_XS · Multiple images at once — provede2b · IQ4_XS · Spatial reasoning — provede2b · IQ4_XS · Writes code — provede2b · IQ4_XS · Reasons about code — provede2b · IQ4_XS · Synthesises sources — provede2b · IQ4_XS · Follows instructions — provede2b · Q4_0 · Tool calling — provede2b · Q4_0 · Tool loop finished — provede2b · Q4_0 · Faithful to tool results — provede2b · Q4_0 · Chained tool calls — provede2b · Q4_0 · Structured output — provede2b · Q4_0 · Sustained output — provede2b · Q4_0 · Extended thinking — provede2b · Q4_0 · Vision — provede2b · Q4_0 · Counts in images — provede2b · Q4_0 · Multiple images at once — provede2b · Q4_0 · Spatial reasoning — provede2b · Q4_0 · Writes code — provede2b · Q4_0 · Reasons about code — provede2b · Q4_0 · Synthesises sources — provede2b · Q4_0 · Follows instructions — proved12b · Q4_K_M · Tool calling — proved12b · Q4_K_M · Tool loop finished — proved12b · Q4_K_M · Faithful to tool results — proved12b · Q4_K_M · Chained tool calls — proved12b · Q4_K_M · Structured output — proved12b · Q4_K_M · Sustained output — proved12b · Q4_K_M · Extended thinking — proved12b · Q4_K_M · Vision — proved12b · Q4_K_M · Counts in images — proved12b · Q4_K_M · Multiple images at once — proved12b · Q4_K_M · Spatial reasoning — proved12b · Q4_K_M · Writes code — proved12b · Q4_K_M · Reasons about code — proved12b · Q4_K_M · Synthesises sources — proved12b · Q4_K_M · Follows instructions — proved26b-a4b · Q8_0 · Tool calling — proved26b-a4b · Q8_0 · Tool loop finished — proved26b-a4b · Q8_0 · Faithful to tool results — proved26b-a4b · Q8_0 · Chained tool calls — proved26b-a4b · Q8_0 · Structured output — proved26b-a4b · Q8_0 · Sustained output — proved26b-a4b · Q8_0 · Extended thinking — proved26b-a4b · Q8_0 · Vision — proved26b-a4b · Q8_0 · Counts in images — proved26b-a4b · Q8_0 · Multiple images at once — proved26b-a4b · Q8_0 · Spatial reasoning — proved26b-a4b · Q8_0 · Writes code — proved26b-a4b · Q8_0 · Reasons about code — proved26b-a4b · Q8_0 · Synthesises sources — proved26b-a4b · Q8_0 · Follows instructions — provede2b · BF16 · Tool calling — provede2b · BF16 · Tool loop finished — provede2b · BF16 · Faithful to tool results — provede2b · BF16 · Chained tool calls — provede2b · BF16 · Structured output — provede2b · BF16 · Sustained output — provede2b · BF16 · Extended thinking — provede2b · BF16 · Vision — provede2b · BF16 · Counts in images — provede2b · BF16 · Multiple images at once — provede2b · BF16 · Spatial reasoning — provede2b · BF16 · Writes code — provede2b · BF16 · Reasons about code — provede2b · BF16 · Synthesises sources — provede2b · BF16 · Follows instructions — provede2b · Q8_0 · Tool calling — provede2b · Q8_0 · Tool loop finished — provede2b · Q8_0 · Faithful to tool results — provede2b · Q8_0 · Chained tool calls — provede2b · Q8_0 · Structured output — provede2b · Q8_0 · Sustained output — provede2b · Q8_0 · Extended thinking — provede2b · Q8_0 · Vision — provede2b · Q8_0 · Counts in images — provede2b · Q8_0 · Multiple images at once — provede2b · Q8_0 · Spatial reasoning — provede2b · Q8_0 · Writes code — provede2b · Q8_0 · Reasons about code — provede2b · Q8_0 · Synthesises sources — provede2b · Q8_0 · Follows instructions — provede4b · Q4_K_M · ollama · Tool calling — provede4b · Q4_K_M · ollama · Tool loop finished — provede4b · Q4_K_M · ollama · Faithful to tool results — provede4b · Q4_K_M · ollama · Chained tool calls — provede4b · Q4_K_M · ollama · Structured output — provede4b · Q4_K_M · ollama · Sustained output — provede4b · Q4_K_M · ollama · Extended thinking — provede4b · Q4_K_M · ollama · Vision — provede4b · Q4_K_M · ollama · Counts in images — provede4b · Q4_K_M · ollama · Multiple images at once — provede4b · Q4_K_M · ollama · Spatial reasoning — provede4b · Q4_K_M · ollama · Writes code — provede4b · Q4_K_M · ollama · Reasons about code — provede4b · Q4_K_M · ollama · Synthesises sources — provede4b · Q4_K_M · ollama · Follows instructions — provede2b~bartowski · BF16 · Tool calling — provede2b~bartowski · BF16 · Tool loop finished — provede2b~bartowski · BF16 · Faithful to tool results — provede2b~bartowski · BF16 · Chained tool calls — provede2b~bartowski · BF16 · Structured output — provede2b~bartowski · BF16 · Sustained output — provede2b~bartowski · BF16 · Extended thinking — provede2b~bartowski · BF16 · Vision — provede2b~bartowski · BF16 · Counts in images — provede2b~bartowski · BF16 · Multiple images at once — provede2b~bartowski · BF16 · Spatial reasoning — provede2b~bartowski · BF16 · Writes code — provede2b~bartowski · BF16 · Reasons about code — provede2b~bartowski · BF16 · Synthesises sources — provede2b~bartowski · BF16 · Follows instructions — provede2b~bartowski · IQ2_M · Tool calling — provede2b~bartowski · IQ2_M · Tool loop finished — provede2b~bartowski · IQ2_M · Faithful to tool results — provede2b~bartowski · IQ2_M · Chained tool calls — provede2b~bartowski · IQ2_M · Structured output — provede2b~bartowski · IQ2_M · Sustained output — provede2b~bartowski · IQ2_M · Extended thinking — provede2b~bartowski · IQ2_M · Vision — provede2b~bartowski · IQ2_M · Counts in images — provede2b~bartowski · IQ2_M · Multiple images at once — provede2b~bartowski · IQ2_M · Spatial reasoning — asked, failede2b~bartowski · IQ2_M · Writes code — provede2b~bartowski · IQ2_M · Reasons about code — provede2b~bartowski · IQ2_M · Synthesises sources — provede2b~bartowski · IQ2_M · Follows instructions — provede2b~bartowski · IQ3_XS · Tool calling — provede2b~bartowski · IQ3_XS · Tool loop finished — provede2b~bartowski · IQ3_XS · Faithful to tool results — provede2b~bartowski · IQ3_XS · Chained tool calls — provede2b~bartowski · IQ3_XS · Structured output — provede2b~bartowski · IQ3_XS · Sustained output — provede2b~bartowski · IQ3_XS · Extended thinking — provede2b~bartowski · IQ3_XS · Vision — provede2b~bartowski · IQ3_XS · Counts in images — provede2b~bartowski · IQ3_XS · Multiple images at once — provede2b~bartowski · IQ3_XS · Spatial reasoning — provede2b~bartowski · IQ3_XS · Writes code — provede2b~bartowski · IQ3_XS · Reasons about code — provede2b~bartowski · IQ3_XS · Synthesises sources — provede2b~bartowski · IQ3_XS · Follows instructions — provede2b~bartowski · Q3_K_L · Tool calling — provede2b~bartowski · Q3_K_L · Tool loop finished — provede2b~bartowski · Q3_K_L · Faithful to tool results — provede2b~bartowski · Q3_K_L · Chained tool calls — provede2b~bartowski · Q3_K_L · Structured output — provede2b~bartowski · Q3_K_L · Sustained output — provede2b~bartowski · Q3_K_L · Extended thinking — provede2b~bartowski · Q3_K_L · Vision — provede2b~bartowski · Q3_K_L · Counts in images — provede2b~bartowski · Q3_K_L · Multiple images at once — provede2b~bartowski · Q3_K_L · Spatial reasoning — provede2b~bartowski · Q3_K_L · Writes code — provede2b~bartowski · Q3_K_L · Reasons about code — provede2b~bartowski · Q3_K_L · Synthesises sources — provede2b~bartowski · Q3_K_L · Follows instructions — provede2b~bartowski · Q4_0 · Tool calling — provede2b~bartowski · Q4_0 · Tool loop finished — provede2b~bartowski · Q4_0 · Faithful to tool results — provede2b~bartowski · Q4_0 · Chained tool calls — provede2b~bartowski · Q4_0 · Structured output — provede2b~bartowski · Q4_0 · Sustained output — provede2b~bartowski · Q4_0 · Extended thinking — provede2b~bartowski · Q4_0 · Vision — provede2b~bartowski · Q4_0 · Counts in images — provede2b~bartowski · Q4_0 · Multiple images at once — provede2b~bartowski · Q4_0 · Spatial reasoning — provede2b~bartowski · Q4_0 · Writes code — provede2b~bartowski · Q4_0 · Reasons about code — provede2b~bartowski · Q4_0 · Synthesises sources — provede2b~bartowski · Q4_0 · Follows instructions — provede2b~bartowski · Q4_K_M · Tool calling — provede2b~bartowski · Q4_K_M · Tool loop finished — provede2b~bartowski · Q4_K_M · Faithful to tool results — provede2b~bartowski · Q4_K_M · Chained tool calls — provede2b~bartowski · Q4_K_M · Structured output — provede2b~bartowski · Q4_K_M · Sustained output — provede2b~bartowski · Q4_K_M · Extended thinking — provede2b~bartowski · Q4_K_M · Vision — provede2b~bartowski · Q4_K_M · Counts in images — provede2b~bartowski · Q4_K_M · Multiple images at once — provede2b~bartowski · Q4_K_M · Spatial reasoning — provede2b~bartowski · Q4_K_M · Writes code — provede2b~bartowski · Q4_K_M · Reasons about code — provede2b~bartowski · Q4_K_M · Synthesises sources — provede2b~bartowski · Q4_K_M · Follows instructions — provede2b~bartowski · Q8_0 · Tool calling — provede2b~bartowski · Q8_0 · Tool loop finished — provede2b~bartowski · Q8_0 · Faithful to tool results — provede2b~bartowski · Q8_0 · Chained tool calls — provede2b~bartowski · Q8_0 · Structured output — provede2b~bartowski · Q8_0 · Sustained output — provede2b~bartowski · Q8_0 · Extended thinking — provede2b~bartowski · Q8_0 · Vision — provede2b~bartowski · Q8_0 · Counts in images — provede2b~bartowski · Q8_0 · Multiple images at once — provede2b~bartowski · Q8_0 · Spatial reasoning — provede2b~bartowski · Q8_0 · Writes code — provede2b~bartowski · Q8_0 · Reasons about code — provede2b~bartowski · Q8_0 · Synthesises sources — provede2b~bartowski · Q8_0 · Follows instructions — proved12b · Q4_0 · Tool calling — proved12b · Q4_0 · Tool loop finished — proved12b · Q4_0 · Faithful to tool results — proved12b · Q4_0 · Chained tool calls — proved12b · Q4_0 · Structured output — proved12b · Q4_0 · Sustained output — proved12b · Q4_0 · Extended thinking — proved12b · Q4_0 · Vision — proved12b · Q4_0 · Counts in images — proved12b · Q4_0 · Multiple images at once — proved12b · Q4_0 · Spatial reasoning — proved12b · Q4_0 · Writes code — proved12b · Q4_0 · Reasons about code — proved12b · Q4_0 · Synthesises sources — proved12b · Q4_0 · Follows instructions — proved12b · BF16 · Tool calling — proved12b · BF16 · Tool loop finished — proved12b · BF16 · Faithful to tool results — proved12b · BF16 · Chained tool calls — proved12b · BF16 · Structured output — proved12b · BF16 · Sustained output — proved12b · BF16 · Extended thinking — proved12b · BF16 · Vision — never asked12b · BF16 · Counts in images — never asked12b · BF16 · Multiple images at once — never asked12b · BF16 · Spatial reasoning — never asked12b · BF16 · Writes code — proved12b · BF16 · Reasons about code — proved12b · BF16 · Synthesises sources — proved12b · BF16 · Follows instructions — proved12b · IQ4_XS · Tool calling — proved12b · IQ4_XS · Tool loop finished — proved12b · IQ4_XS · Faithful to tool results — proved12b · IQ4_XS · Chained tool calls — proved12b · IQ4_XS · Structured output — proved12b · IQ4_XS · Sustained output — proved12b · IQ4_XS · Extended thinking — proved12b · IQ4_XS · Vision — proved12b · IQ4_XS · Counts in images — proved12b · IQ4_XS · Multiple images at once — proved12b · IQ4_XS · Spatial reasoning — proved12b · IQ4_XS · Writes code — proved12b · IQ4_XS · Reasons about code — proved12b · IQ4_XS · Synthesises sources — proved12b · IQ4_XS · Follows instructions — proved31b · Q4_0 · Tool calling — proved31b · Q4_0 · Tool loop finished — proved31b · Q4_0 · Faithful to tool results — proved31b · Q4_0 · Chained tool calls — proved31b · Q4_0 · Structured output — proved31b · Q4_0 · Sustained output — proved31b · Q4_0 · Extended thinking — proved31b · Q4_0 · Vision — proved31b · Q4_0 · Counts in images — proved31b · Q4_0 · Multiple images at once — proved31b · Q4_0 · Spatial reasoning — proved31b · Q4_0 · Writes code — proved31b · Q4_0 · Reasons about code — proved31b · Q4_0 · Synthesises sources — proved31b · Q4_0 · Follows instructions — proved31b · Q4_K_M · Tool calling — proved31b · Q4_K_M · Tool loop finished — proved31b · Q4_K_M · Faithful to tool results — proved31b · Q4_K_M · Chained tool calls — proved31b · Q4_K_M · Structured output — proved31b · Q4_K_M · Sustained output — proved31b · Q4_K_M · Extended thinking — proved31b · Q4_K_M · Vision — proved31b · Q4_K_M · Counts in images — proved31b · Q4_K_M · Multiple images at once — proved31b · Q4_K_M · Spatial reasoning — proved31b · Q4_K_M · Writes code — proved31b · Q4_K_M · Reasons about code — proved31b · Q4_K_M · Synthesises sources — proved31b · Q4_K_M · Follows instructions — provede4b · BF16 · Tool calling — provede4b · BF16 · Tool loop finished — provede4b · BF16 · Faithful to tool results — provede4b · BF16 · Chained tool calls — provede4b · BF16 · Structured output — provede4b · BF16 · Sustained output — provede4b · BF16 · Extended thinking — provede4b · BF16 · Vision — provede4b · BF16 · Counts in images — provede4b · BF16 · Multiple images at once — provede4b · BF16 · Spatial reasoning — provede4b · BF16 · Writes code — provede4b · BF16 · Reasons about code — provede4b · BF16 · Synthesises sources — provede4b · BF16 · Follows instructions — provede4b · IQ4_XS · Tool calling — provede4b · IQ4_XS · Tool loop finished — provede4b · IQ4_XS · Faithful to tool results — provede4b · IQ4_XS · Chained tool calls — provede4b · IQ4_XS · Structured output — provede4b · IQ4_XS · Sustained output — provede4b · IQ4_XS · Extended thinking — provede4b · IQ4_XS · Vision — provede4b · IQ4_XS · Counts in images — provede4b · IQ4_XS · Multiple images at once — provede4b · IQ4_XS · Spatial reasoning — provede4b · IQ4_XS · Writes code — provede4b · IQ4_XS · Reasons about code — provede4b · IQ4_XS · Synthesises sources — provede4b · IQ4_XS · Follows instructions — provede4b · Q4_0 · Tool calling — provede4b · Q4_0 · Tool loop finished — provede4b · Q4_0 · Faithful to tool results — provede4b · Q4_0 · Chained tool calls — provede4b · Q4_0 · Structured output — provede4b · Q4_0 · Sustained output — provede4b · Q4_0 · Extended thinking — provede4b · Q4_0 · Vision — provede4b · Q4_0 · Counts in images — provede4b · Q4_0 · Multiple images at once — provede4b · Q4_0 · Spatial reasoning — provede4b · Q4_0 · Writes code — provede4b · Q4_0 · Reasons about code — provede4b · Q4_0 · Synthesises sources — provede4b · Q4_0 · Follows instructions — provede4b · Q4_K_M · unsloth · Tool calling — provede4b · Q4_K_M · unsloth · Tool loop finished — provede4b · Q4_K_M · unsloth · Faithful to tool results — provede4b · Q4_K_M · unsloth · Chained tool calls — provede4b · Q4_K_M · unsloth · Structured output — provede4b · Q4_K_M · unsloth · Sustained output — provede4b · Q4_K_M · unsloth · Extended thinking — provede4b · Q4_K_M · unsloth · Vision — provede4b · Q4_K_M · unsloth · Counts in images — provede4b · Q4_K_M · unsloth · Multiple images at once — provede4b · Q4_K_M · unsloth · Spatial reasoning — asked, failede4b · Q4_K_M · unsloth · Writes code — provede4b · Q4_K_M · unsloth · Reasons about code — provede4b · Q4_K_M · unsloth · Synthesises sources — provede4b · Q4_K_M · unsloth · Follows instructions — provede4b · Q8_0 · Tool calling — provede4b · Q8_0 · Tool loop finished — asked, failede4b · Q8_0 · Faithful to tool results — provede4b · Q8_0 · Chained tool calls — provede4b · Q8_0 · Structured output — provede4b · Q8_0 · Sustained output — provede4b · Q8_0 · Extended thinking — provede4b · Q8_0 · Vision — provede4b · Q8_0 · Counts in images — provede4b · Q8_0 · Multiple images at once — provede4b · Q8_0 · Spatial reasoning — provede4b · Q8_0 · Writes code — provede4b · Q8_0 · Reasons about code — provede4b · Q8_0 · Synthesises sources — provede4b · Q8_0 · Follows instructions — proved12b~yuxinlu1 · Q3_K_M · Tool calling — proved12b~yuxinlu1 · Q3_K_M · Tool loop finished — proved12b~yuxinlu1 · Q3_K_M · Faithful to tool results — asked, failed12b~yuxinlu1 · Q3_K_M · Chained tool calls — never asked12b~yuxinlu1 · Q3_K_M · Structured output — proved12b~yuxinlu1 · Q3_K_M · Sustained output — proved12b~yuxinlu1 · Q3_K_M · Extended thinking — proved12b~yuxinlu1 · Q3_K_M · Vision — never asked12b~yuxinlu1 · Q3_K_M · Counts in images — never asked12b~yuxinlu1 · Q3_K_M · Multiple images at once — never asked12b~yuxinlu1 · Q3_K_M · Spatial reasoning — never asked12b~yuxinlu1 · Q3_K_M · Writes code — proved12b~yuxinlu1 · Q3_K_M · Reasons about code — proved12b~yuxinlu1 · Q3_K_M · Synthesises sources — proved12b~yuxinlu1 · Q3_K_M · Follows instructions — proved12b~yuxinlu1 · Q4_K_M · Tool calling — proved12b~yuxinlu1 · Q4_K_M · Tool loop finished — asked, failed12b~yuxinlu1 · Q4_K_M · Faithful to tool results — proved12b~yuxinlu1 · Q4_K_M · Chained tool calls — proved12b~yuxinlu1 · Q4_K_M · Structured output — proved12b~yuxinlu1 · Q4_K_M · Sustained output — proved12b~yuxinlu1 · Q4_K_M · Extended thinking — proved12b~yuxinlu1 · Q4_K_M · Vision — never asked12b~yuxinlu1 · Q4_K_M · Counts in images — never asked12b~yuxinlu1 · Q4_K_M · Multiple images at once — never asked12b~yuxinlu1 · Q4_K_M · Spatial reasoning — never asked12b~yuxinlu1 · Q4_K_M · Writes code — proved12b~yuxinlu1 · Q4_K_M · Reasons about code — proved12b~yuxinlu1 · Q4_K_M · Synthesises sources — proved12b~yuxinlu1 · Q4_K_M · Follows instructions — proved12b~yuxinlu1 · Q6_K · Tool calling — proved12b~yuxinlu1 · Q6_K · Tool loop finished — asked, failed12b~yuxinlu1 · Q6_K · Faithful to tool results — proved12b~yuxinlu1 · Q6_K · Chained tool calls — proved12b~yuxinlu1 · Q6_K · Structured output — proved12b~yuxinlu1 · Q6_K · Sustained output — proved12b~yuxinlu1 · Q6_K · Extended thinking — proved12b~yuxinlu1 · Q6_K · Vision — never asked12b~yuxinlu1 · Q6_K · Counts in images — never asked12b~yuxinlu1 · Q6_K · Multiple images at once — never asked12b~yuxinlu1 · Q6_K · Spatial reasoning — never asked12b~yuxinlu1 · Q6_K · Writes code — proved12b~yuxinlu1 · Q6_K · Reasons about code — proved12b~yuxinlu1 · Q6_K · Synthesises sources — proved12b~yuxinlu1 · Q6_K · Follows instructions — proved12b~yuxinlu1 · Q8_0 · Tool calling — proved12b~yuxinlu1 · Q8_0 · Tool loop finished — asked, failed12b~yuxinlu1 · Q8_0 · Faithful to tool results — proved12b~yuxinlu1 · Q8_0 · Chained tool calls — proved12b~yuxinlu1 · Q8_0 · Structured output — proved12b~yuxinlu1 · Q8_0 · Sustained output — proved12b~yuxinlu1 · Q8_0 · Extended thinking — proved12b~yuxinlu1 · Q8_0 · Vision — never asked12b~yuxinlu1 · Q8_0 · Counts in images — never asked12b~yuxinlu1 · Q8_0 · Multiple images at once — never asked12b~yuxinlu1 · Q8_0 · Spatial reasoning — never asked12b~yuxinlu1 · Q8_0 · Writes code — proved12b~yuxinlu1 · Q8_0 · Reasons about code — proved12b~yuxinlu1 · Q8_0 · Synthesises sources — proved12b~yuxinlu1 · Q8_0 · Follows instructions — provede2b · Q3_K_M · Tool calling — provede2b · Q3_K_M · Tool loop finished — asked, failede2b · Q3_K_M · Faithful to tool results — provede2b · Q3_K_M · Chained tool calls — provede2b · Q3_K_M · Structured output — provede2b · Q3_K_M · Sustained output — provede2b · Q3_K_M · Extended thinking — provede2b · Q3_K_M · Vision — provede2b · Q3_K_M · Counts in images — provede2b · Q3_K_M · Multiple images at once — provede2b · Q3_K_M · Spatial reasoning — provede2b · Q3_K_M · Writes code — provede2b · Q3_K_M · Reasons about code — provede2b · Q3_K_M · Synthesises sources — provede2b · Q3_K_M · Follows instructions — provede2b · IQ4_NL · Tool calling — provede2b · IQ4_NL · Tool loop finished — provede2b · IQ4_NL · Faithful to tool results — provede2b · IQ4_NL · Chained tool calls — provede2b · IQ4_NL · Structured output — provede2b · IQ4_NL · Sustained output — provede2b · IQ4_NL · Extended thinking — provede2b · IQ4_NL · Vision — provede2b · IQ4_NL · Counts in images — provede2b · IQ4_NL · Multiple images at once — provede2b · IQ4_NL · Spatial reasoning — provede2b · IQ4_NL · Writes code — provede2b · IQ4_NL · Reasons about code — provede2b · IQ4_NL · Synthesises sources — provede2b · IQ4_NL · Follows instructions — provede2b · Q3_K_S · Tool calling — provede2b · Q3_K_S · Tool loop finished — asked, failede2b · Q3_K_S · Faithful to tool results — provede2b · Q3_K_S · Chained tool calls — provede2b · Q3_K_S · Structured output — provede2b · Q3_K_S · Sustained output — provede2b · Q3_K_S · Extended thinking — provede2b · Q3_K_S · Vision — provede2b · Q3_K_S · Counts in images — provede2b · Q3_K_S · Multiple images at once — provede2b · Q3_K_S · Spatial reasoning — provede2b · Q3_K_S · Writes code — provede2b · Q3_K_S · Reasons about code — provede2b · Q3_K_S · Synthesises sources — provede2b · Q3_K_S · Follows instructions — provede4b · Q3_K_M · Tool calling — provede4b · Q3_K_M · Tool loop finished — provede4b · Q3_K_M · Faithful to tool results — provede4b · Q3_K_M · Chained tool calls — provede4b · Q3_K_M · Structured output — provede4b · Q3_K_M · Sustained output — provede4b · Q3_K_M · Extended thinking — provede4b · Q3_K_M · Vision — provede4b · Q3_K_M · Counts in images — provede4b · Q3_K_M · Multiple images at once — provede4b · Q3_K_M · Spatial reasoning — provede4b · Q3_K_M · Writes code — provede4b · Q3_K_M · Reasons about code — provede4b · Q3_K_M · Synthesises sources — provede4b · Q3_K_M · Follows instructions — provede2b · e2b-it-qat · Tool calling — never askede2b · e2b-it-qat · Tool loop finished — never askede2b · e2b-it-qat · Faithful to tool results — never askede2b · e2b-it-qat · Chained tool calls — never askede2b · e2b-it-qat · Structured output — never askede2b · e2b-it-qat · Sustained output — never askede2b · e2b-it-qat · Extended thinking — never askede2b · e2b-it-qat · Vision — never askede2b · e2b-it-qat · Counts in images — never askede2b · e2b-it-qat · Multiple images at once — never askede2b · e2b-it-qat · Spatial reasoning — never askede2b · e2b-it-qat · Writes code — never askede2b · e2b-it-qat · Reasons about code — never askede2b · e2b-it-qat · Synthesises sources — never askede2b · e2b-it-qat · Follows instructions — never askede4b · e4b-it-qat · Tool calling — never askede4b · e4b-it-qat · Tool loop finished — never askede4b · e4b-it-qat · Faithful to tool results — never askede4b · e4b-it-qat · Chained tool calls — never askede4b · e4b-it-qat · Structured output — never askede4b · e4b-it-qat · Sustained output — never askede4b · e4b-it-qat · Extended thinking — never askede4b · e4b-it-qat · Vision — never askede4b · e4b-it-qat · Counts in images — never askede4b · e4b-it-qat · Multiple images at once — never askede4b · e4b-it-qat · Spatial reasoning — never askede4b · e4b-it-qat · Writes code — never askede4b · e4b-it-qat · Reasons about code — never askede4b · e4b-it-qat · Synthesises sources — never askede4b · e4b-it-qat · Follows instructions — never askede2b · Tool calling — never askede2b · Tool loop finished — never askede2b · Faithful to tool results — never askede2b · Chained tool calls — never askede2b · Structured output — never askede2b · Sustained output — never askede2b · Extended thinking — never askede2b · Vision — never askede2b · Counts in images — never askede2b · Multiple images at once — never askede2b · Spatial reasoning — never askede2b · Writes code — never askede2b · Reasons about code — never askede2b · Synthesises sources — never askede2b · Follows instructions — never askede2b · Q2_K_XL · Tool calling — provede2b · Q2_K_XL · Tool loop finished — asked, failede2b · Q2_K_XL · Faithful to tool results — asked, failede2b · Q2_K_XL · Chained tool calls — never askede2b · Q2_K_XL · Structured output — provede2b · Q2_K_XL · Sustained output — provede2b · Q2_K_XL · Extended thinking — provede2b · Q2_K_XL · Vision — never askede2b · Q2_K_XL · Counts in images — never askede2b · Q2_K_XL · Multiple images at once — never askede2b · Q2_K_XL · Spatial reasoning — never askede2b · Q2_K_XL · Writes code — provede2b · Q2_K_XL · Reasons about code — never askede2b · Q2_K_XL · Synthesises sources — provede2b · Q2_K_XL · Follows instructions — proved
Tool calling
Tool loop finished
Faithful to tool results
Chained tool calls
Structured output
Sustained output
Extended thinking
Vision
Counts in images
Multiple images at once
Spatial reasoning
Writes code
Reasons about code
Synthesises sources
Follows instructions
Gemma 4 12B5 builds measured →
hf.co/unsloth/gemma-4-12b-it-GGUF:Q8_012b · Q8_0 · unsloth · ollama · 1 machine · 11.9B
256Krecalled
—served
18.6best tok/s
recalled 256k⌐ declared 256k · unsigned
hf.co/unsloth/gemma-4-12b-it-GGUF:Q4_K_M12b · Q4_K_M · unsloth · ollama · 2 machines · 11.9B
128Krecalled
—served
29.1best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
hf.co/google/gemma-4-12B-it-qat-q4_0-gguf:Q4_012b · Q4_0 · google · ollama · 1 machine · 11.9B
128Krecalled
—served
27.9best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
hf.co/unsloth/gemma-4-12b-it-GGUF:BF1612b · BF16 · unsloth · ollama · 1 machine · 11.9B
128Krecalled
—served
10.7best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
hf.co/unsloth/gemma-4-12b-it-GGUF:IQ4_XS12b · IQ4_XS · unsloth · ollama · 1 machine · 11.9B
128Krecalled
—served
23.2best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
Gemma 4 E2B13 builds measured →
hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_me2b · Q4_K_M · unsloth · ollama · 13 machines · 4.65B
128Krecalled
32Kserved
291.2best tok/s
291.2 tok/s · unnamed machine98.9 tok/s · unnamed machine87.4 tok/s · unnamed machine77.6 tok/s · unnamed machine74.4 tok/s · unnamed machine74.2 tok/s · unnamed machine53.7 tok/s · unnamed machine28.1 tok/s · unnamed machine24.9 tok/s · unnamed machine15.3 tok/s · unnamed machine10.0 tok/s · unnamed machine6.8 tok/s · x86 · CPU · Bengaluru0.7 tok/s · Apple silicon · Bhubaneswar
0.7 → 291.2 tok/s across 13 machines · ×432.8 spreadunnamed machinex86 · CPUApple silicon
recalled 128k▲ served 32k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_Me2b · Q4_K_M · unsloth · ollama · 13 machines · 4.65B
128Krecalled
32Kserved
291.2best tok/s
291.2 tok/s · unnamed machine98.9 tok/s · unnamed machine87.4 tok/s · unnamed machine77.6 tok/s · unnamed machine74.4 tok/s · unnamed machine74.2 tok/s · unnamed machine53.7 tok/s · unnamed machine28.1 tok/s · unnamed machine24.9 tok/s · unnamed machine15.3 tok/s · unnamed machine10.0 tok/s · unnamed machine6.8 tok/s · x86 · CPU · Bengaluru4.0 tok/s · Apple silicon · Bhubaneswar
4.0 → 291.2 tok/s across 13 machines · ×73.0 spreadunnamed machinex86 · CPUApple silicon
recalled 128k▲ served 32k⌐ declared 128k · unsigned
gemma4:e2be2b · Q4_K_M · google · ollama · 10 machines · 5.1B
128Krecalled
32Kserved
170.5best tok/s
170.5 tok/s · unnamed machine149.9 tok/s · unnamed machine91.1 tok/s · unnamed machine43.3 tok/s · unnamed machine19.3 tok/s · unnamed machine19.3 tok/s · unnamed machine8.9 tok/s · x86 · CPU · Hyderabad7.7 tok/s · x86 · CPU · Bengaluru7.1 tok/s · unnamed machine4.1 tok/s · Apple silicon · Bhubaneswar
4.1 → 170.5 tok/s across 10 machines · ×42.0 spreadunnamed machinex86 · CPUApple silicon
recalled 128k▲ served 32k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XSe2b · IQ4_XS · unsloth · ollama · 3 machines · 4.65B
128Krecalled
—served
182.9best tok/s
182.9 tok/s · unnamed machine78.7 tok/s · unnamed machine6.9 tok/s · unnamed machine
6.9 → 182.9 tok/s across 3 machines · ×26.4 spreadunnamed machine
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0e2b · Q4_0 · unsloth · ollama · 3 machines · 4.65B
128Krecalled
—served
84.7best tok/s
83.8 tok/s · unnamed machine57.6 tok/s · unnamed machine9.1 tok/s · x86 · CPU · Bengaluru
9.1 → 83.8 tok/s across 3 machines · ×9.2 spreadunnamed machinex86 · CPU
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E2B-it-GGUF:BF16e2b · BF16 · unsloth · ollama · 2 machines · 4.65B
128Krecalled
—served
38.3best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E2B-it-GGUF:Q8_0e2b · Q8_0 · unsloth · ollama · 2 machines · 4.65B
128Krecalled
—served
67.0best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_Me2b · Q3_K_M · unsloth · ollama · 3 machines · 4.65B
8Krecalled
—served
38.4best tok/s
38.4 tok/s · unnamed machine14.2 tok/s · x86 · CPU · Bengaluru2.8 tok/s · Apple silicon · Bhubaneswar
2.8 → 38.4 tok/s across 3 machines · ×13.8 spreadunnamed machinex86 · CPUApple silicon
recalled 8k⌐ declared 128k · unsignedadvertises 16× what it proved
hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_NLe2b · IQ4_NL · unsloth · ollama · 2 machines · 4.65B
8Krecalled
—served
54.9best tok/s
54.9 tok/s · unnamed machine10.0 tok/s · x86 · CPU · Bengaluru
10.0 → 54.9 tok/s across 2 machines · ×5.5 spreadunnamed machinex86 · CPU
recalled 8k⌐ declared 128k · unsignedadvertises 16× what it proved
hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_Se2b · Q3_K_S · unsloth · ollama · 1 machine · 4.65B
4Krecalled
—served
14.4best tok/s
recalled 4k⌐ declared 128k · unsignedadvertises 32× what it proved
gemma4:e2b-it-qate2b · e2b-it-qat · ollama · ollama · 1 machine
—recalled
—served
8.9best tok/s
google/gemma-4-E2B-it-qat-q4_0-ggufe2b · google · llamacpp · 1 machine
—recalled
—served
2.0best tok/s
unsloth/gemma-4-E2B-it-qat-mobile-GGUF:Q2_K_XLe2b · Q2_K_XL · unsloth · llamacpp · 1 machine
—recalled
—served
—best tok/s
Gemma 4 26B-A4B1 build measured →
hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q8_026b-a4b · Q8_0 · unsloth · ollama · 2 machines · 25.2B
128Krecalled
—served
20.8best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
Gemma 4 E4B8 builds measured →
gemma4:e4be4b · Q4_K_M · google · ollama · 1 machine · 7.5B
128Krecalled
—served
93.7best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E4B-it-GGUF:BF16e4b · BF16 · unsloth · ollama · 1 machine · 7.52B
128Krecalled
—served
25.8best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ4_XSe4b · IQ4_XS · unsloth · ollama · 1 machine · 7.52B
128Krecalled
—served
61.4best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_0e4b · Q4_0 · unsloth · ollama · 1 machine · 7.52B
128Krecalled
—served
54.7best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_K_Me4b · Q4_K_M · unsloth · ollama · 1 machine · 7.52B
128Krecalled
—served
56.8best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E4B-it-GGUF:Q8_0e4b · Q8_0 · unsloth · ollama · 1 machine · 7.52B
128Krecalled
—served
39.5best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gemma-4-E4B-it-GGUF:Q3_K_Me4b · Q3_K_M · unsloth · ollama · 1 machine
4Krecalled
—served
20.4best tok/s
gemma4:e4b-it-qate4b · e4b-it-qat · ollama · ollama · 1 machine
—recalled
—served
—best tok/s
e2b~bartowski7 builds measured →
hf.co/bartowski/google_gemma-4-E2B-it-GGUF:BF16e2b~bartowski · BF16 · bartowski · ollama · 1 machine · 4.65B
128Krecalled
—served
28.2best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/bartowski/google_gemma-4-E2B-it-GGUF:IQ2_Me2b~bartowski · IQ2_M · bartowski · ollama · 1 machine · 4.65B
128Krecalled
—served
71.2best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/bartowski/google_gemma-4-E2B-it-GGUF:IQ3_XSe2b~bartowski · IQ3_XS · bartowski · ollama · 1 machine · 4.65B
128Krecalled
—served
62.3best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q3_K_Le2b~bartowski · Q3_K_L · bartowski · ollama · 1 machine · 4.65B
128Krecalled
—served
73.0best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q4_0e2b~bartowski · Q4_0 · bartowski · ollama · 1 machine · 4.65B
128Krecalled
—served
87.3best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q4_K_Me2b~bartowski · Q4_K_M · bartowski · ollama · 1 machine · 4.65B
128Krecalled
—served
73.4best tok/s
recalled 128k⌐ declared 128k · unsigned
hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q8_0e2b~bartowski · Q8_0 · bartowski · ollama · 1 machine · 4.65B
128Krecalled
—served
61.3best tok/s
recalled 128k⌐ declared 128k · unsigned
Gemma 4 31B2 builds measured →
hf.co/unsloth/gemma-4-31B-it-GGUF:Q4_031b · Q4_0 · unsloth · ollama · 1 machine · 30.7B
128Krecalled
—served
2.4best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
hf.co/unsloth/gemma-4-31B-it-GGUF:Q4_K_M31b · Q4_K_M · unsloth · ollama · 1 machine · 30.7B
128Krecalled
—served
3.4best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
12b~yuxinlu14 builds measured →
hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q3_K_M12b~yuxinlu1 · Q3_K_M · yuxinlu1 · ollama · 1 machine · 11.9B
128Krecalled
—served
17.0best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q4_K_M12b~yuxinlu1 · Q4_K_M · yuxinlu1 · ollama · 1 machine · 11.9B
128Krecalled
—served
25.8best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q6_K12b~yuxinlu1 · Q6_K · yuxinlu1 · ollama · 1 machine · 11.9B
128Krecalled
—served
20.0best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved
hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q8_012b~yuxinlu1 · Q8_0 · yuxinlu1 · ollama · 1 machine · 11.9B
128Krecalled
—served
18.0best tok/s
recalled 128k⌐ declared 256k · unsignedadvertises 2× what it proved

Builds this network knows of

Declared, not measured — this is what the catalog says exists and what it would cost a machine to hold. Nothing here is signed, and none of it claims the build works.

how much memory does your machine have

48 of 69 builds fit a 16 GB machine — budgeting 70% of it, or 11.2 GiB, because the cache and the runtime want the rest. 2 of them use a signed resident size a machine actually held; the rest are declared file sizes.

  • google/gemma-4-26B-A4B-it-assistantBF16vllmpublished by google
    0.8 GiB
  • google/gemma-4-12B-it-assistantBF16vllmpublished by google
    0.8 GiB
  • hf.co/turintech/gemma-4-E2B-it-Artemis-GGUF:IQ2_MIQ2_Mollamaderived by turintech
    1.7 GiB
  • hf.co/turintech/gemma-4-E2B-it-Artemis-GGUF:IQ3_XXSIQ3_XXSollamaderived by turintech
    1.9 GiB
  • hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ2_MIQ2_Mollamaquantized by unsloth
    2.1 GiB
  • hf.co/turintech/gemma-4-E2B-it-Artemis-GGUF:Q3_K_LQ3_K_Lollamaderived by turintech
    2.1 GiB
  • hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ3_XXSIQ3_XXSollamaquantized by unsloth
    2.2 GiB
  • hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XSIQ4_XSollamameasured herequantized by unsloth
    2.8 GiB
  • mlx-community/gemma-4-E2B-it-4bit4bitmlxquantized by mlx-community
    2.8 GiB
  • hf.co/google/gemma-4-E2B-it-qat-q4_0-gguf:Q4_0Q4_0ollamapublished by google
    3.1 GiB
  • hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ2_MIQ2_Mollamaquantized by unsloth
    3.3 GiB
  • hf.co/turintech/gemma-4-E4B-it-Artemis-GGUF:Q3_K_MQ3_K_Mollamaderived by turintech
    3.3 GiB
  • hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ3_XXSIQ3_XXSollamaquantized by unsloth
    3.5 GiB
  • hf.co/unsloth/gemma-4-12b-it-GGUF:IQ2_MIQ2_Mollamaquantized by unsloth
    3.9 GiB
  • hf.co/turintech/gemma-4-E4B-it-Artemis-GGUF:Q4_0Q4_0ollamaderived by turintech
    3.9 GiB
  • lmstudio-community/gemma-4-E2B-it-MLX-4bit4BITmlxquantized by lmstudio-community
    4.0 GiB
  • hf.co/unsloth/gemma-4-12b-it-GGUF:IQ3_XXSIQ3_XXSollamaquantized by unsloth
    4.3 GiB
  • hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ4_XSIQ4_XSollamameasured herequantized by unsloth
    4.4 GiB
  • lmstudio-community/gemma-4-E2B-it-MLX-5bit5BITmlxquantized by lmstudio-community
    4.4 GiB
  • mlx-community/gemma-4-E4B-it-4bit4bitmlxquantized by mlx-community
    4.5 GiB
  • hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_K_MQ4_K_Mollamameasured herequantized by unsloth
    4.6 GiB
  • hf.co/unsloth/gemma-4-E2B-it-GGUF:Q8_0Q8_0ollamameasured herequantized by unsloth
    4.7 GiB
  • hf.co/google/gemma-4-E4B-it-qat-q4_0-gguf:Q4_0Q4_0ollamapublished by google
    4.8 GiB
  • lmstudio-community/gemma-4-E2B-it-MLX-8bit8BITmlxquantized by lmstudio-community
    5.5 GiB
  • hf.co/unsloth/gemma-4-12b-it-GGUF:IQ4_XSIQ4_XSollamameasured herequantized by unsloth
    5.9 GiB
  • lmstudio-community/gemma-4-12B-it-MLX-4bit4BITmlxquantized by lmstudio-community
    6.3 GiB
  • lmstudio-community/gemma-4-E4B-it-MLX-4bit4BITmlxquantized by lmstudio-community
    6.4 GiB
  • hf.co/google/gemma-4-12B-it-qat-q4_0-gguf:Q4_0Q4_0ollamameasured herepublished by google
    6.5 GiB
  • hf.co/unsloth/gemma-4-12b-it-GGUF:Q4_K_MQ4_K_Mollamameasured herequantized by unsloth
    6.6 GiB
  • mlx-community/gemma-4-12b-it-4bit4bitmlxquantized by mlx-community
    6.8 GiB
  • hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_MQ4_K_Mollamaresident · signedquantized by unsloth
    7.0 GiB
  • hf.co/unsloth/gemma-4-12B-it-qat-GGUF:Q4_K_MQ4_K_Mollamaquantized by unsloth
    7.2 GiB
  • lmstudio-community/gemma-4-E4B-it-MLX-6bit6BITmlxquantized by lmstudio-community
    7.3 GiB
  • hf.co/unsloth/gemma-4-E4B-it-GGUF:Q8_0Q8_0ollamameasured herequantized by unsloth
    7.6 GiB
  • google/gemma-4-E2B-it-qat-w4a16-ctW4A16vllmpublished by google
    7.8 GiB
  • hf.co/unsloth/gemma-4-31B-it-GGUF:IQ2_XXSIQ2_XXSollamaquantized by unsloth
    7.9 GiB
  • lmstudio-community/gemma-4-E4B-it-MLX-8bit8BITmlxquantized by lmstudio-community
    8.3 GiB
  • hf.co/unsloth/gemma-4-E2B-it-GGUF:BF16BF16ollamameasured herequantized by unsloth
    8.7 GiB
  • lmstudio-community/gemma-4-12B-it-MLX-6bit6BITmlxquantized by lmstudio-community
    9.1 GiB
  • hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:IQ2_XXSIQ2_XXSollamaquantized by unsloth
    9.2 GiB
  • hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:IQ2_MIQ2_Mollamaquantized by unsloth
    9.3 GiB
  • google/gemma-4-12B-it-qat-w4a16-ctW4A16vllmpublished by google
    9.6 GiB
  • hf.co/unsloth/gemma-4-31B-it-GGUF:IQ2_MIQ2_Mollamaquantized by unsloth
    10.0 GiB
  • gemma4:e2bQ4_K_Mollamaresident · signedpublished by google
    10.2 GiB
  • hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:IQ3_XXSIQ3_XXSollamaquantized by unsloth
    10.6 GiB
  • google/gemma-4-E4B-it-qat-w4a16-ctW4A16vllmpublished by google
    10.7 GiB
  • gemma4:e4bQ4_K_Mollamameasured herepublished by google
    10.7 GiB
  • hf.co/unsloth/gemma-4-31B-it-GGUF:IQ3_XXSIQ3_XXSollamaquantized by unsloth
    11.0 GiB
  • hf.co/unsloth/gemma-4-12b-it-GGUF:Q8_0Q8_0ollamameasured herequantized by unsloth
    11.8 GiB
  • lmstudio-community/gemma-4-12B-it-MLX-8bit8BITmlxquantized by lmstudio-community
    11.8 GiB
  • hf.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf:Q4_0Q4_0ollamapublished by google
    13.4 GiB
  • hf.co/unsloth/gemma-4-E4B-it-GGUF:BF16BF16ollamameasured herequantized by unsloth
    14.0 GiB
  • lmstudio-community/gemma-4-26B-A4B-it-QAT-MLX-4bit4BITmlxquantized by lmstudio-community
    14.5 GiB
  • hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q4_K_MQ4_K_Mollamaquantized by unsloth
    15.8 GiB
  • cyankiwi/gemma-4-26B-A4B-it-AWQ-4bitW4A16vllmquantized by cyankiwi
    16.0 GiB
  • hf.co/google/gemma-4-31B-it-qat-q4_0-gguf:Q4_0Q4_0ollamapublished by google
    16.4 GiB
  • hf.co/unsloth/gemma-4-31B-it-GGUF:Q4_K_MQ4_K_Mollamameasured herequantized by unsloth
    17.1 GiB
  • nvidia/Gemma-4-26B-A4B-NVFP4NVFP4vllmquantized by nvidia
    17.5 GiB
  • google/gemma-4-31B-it-qat-w4a16-ctW4A16vllmpublished by google
    21.7 GiB
  • hf.co/unsloth/gemma-4-12b-it-GGUF:BF16BF16ollamameasured herequantized by unsloth
    22.2 GiB
  • google/gemma-4-12B-itBF16vllmpublished by google
    22.3 GiB
  • hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q8_0Q8_0ollamameasured herequantized by unsloth
    25.0 GiB
  • RedHatAI/gemma-4-26B-A4B-it-FP8-dynamicFP8vllmquantized by RedHatAI
    26.7 GiB
  • nvidia/Gemma-4-31B-IT-NVFP4NVFP4vllmquantized by nvidia
    30.4 GiB
  • hf.co/unsloth/gemma-4-31B-it-GGUF:Q8_0Q8_0ollamaquantized by unsloth
    30.4 GiB
  • unsloth/gemma-4-26B-A4B-it-GGUF:BF16BF16llamacppquantized by unsloth
    47.0 GiB
  • hf.co/ggml-org/gemma-4-26B-A4B-it-GGUF:BF16BF16ollamaquantized by ggml-org
    47.0 GiB
  • unsloth/gemma-4-31B-it-GGUF:BF16BF16llamacppquantized by unsloth
    57.2 GiB
  • google/gemma-4-31B-it-qat-q4_0-unquantizedBF16vllmpublished by google
    58.3 GiB

Sizes are what the catalog declares, not what a machine measured — fitting on disk is necessary and not sufficient. Nobody here has run these on your machine, and this narrows the candidates rather than certifying one.

the same builds, by member
Gemma 4 12B12B
hf.co/unsloth/gemma-4-12b-it-GGUF:IQ2_MIQ2_Mollama3.9 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-12b-it-GGUF:IQ3_XXSIQ3_XXSollama4.3 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-12b-it-GGUF:IQ4_XSIQ4_XSollama5.9 GiB on diskquantized by unsloth
lmstudio-community/gemma-4-12B-it-MLX-4bit4BITmlx6.3 GiB on diskquantized by lmstudio-community
hf.co/google/gemma-4-12B-it-qat-q4_0-gguf:Q4_0Q4_0ollama6.5 GiB on diskpublished by google
hf.co/unsloth/gemma-4-12b-it-GGUF:Q4_K_MQ4_K_Mollama6.6 GiB on diskquantized by unsloth
mlx-community/gemma-4-12b-it-4bit4bitmlx6.8 GiB on diskquantized by mlx-community
hf.co/unsloth/gemma-4-12B-it-qat-GGUF:Q4_K_MQ4_K_Mollama7.2 GiB on diskquantized by unsloth
lmstudio-community/gemma-4-12B-it-MLX-6bit6BITmlx9.1 GiB on diskquantized by lmstudio-community
google/gemma-4-12B-it-qat-w4a16-ctW4A16vllm9.6 GiB on diskpublished by google
hf.co/unsloth/gemma-4-12b-it-GGUF:Q8_0Q8_0ollama11.8 GiB on diskquantized by unsloth
lmstudio-community/gemma-4-12B-it-MLX-8bit8BITmlx11.8 GiB on diskquantized by lmstudio-community
hf.co/unsloth/gemma-4-12b-it-GGUF:BF16BF16ollama22.2 GiB on diskquantized by unsloth
google/gemma-4-12B-itBF16vllm22.3 GiB on diskpublished by google
Gemma 4 26B-A4B
hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:IQ2_XXSIQ2_XXSollama9.2 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:IQ2_MIQ2_Mollama9.3 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:IQ3_XXSIQ3_XXSollama10.6 GiB on diskquantized by unsloth
hf.co/google/gemma-4-26B-A4B-it-qat-q4_0-gguf:Q4_0Q4_0ollama13.4 GiB on diskpublished by google
lmstudio-community/gemma-4-26B-A4B-it-QAT-MLX-4bit4BITmlx14.5 GiB on diskquantized by lmstudio-community
hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q4_K_MQ4_K_Mollama15.8 GiB on diskquantized by unsloth
cyankiwi/gemma-4-26B-A4B-it-AWQ-4bitW4A16vllm16.0 GiB on diskquantized by cyankiwi
nvidia/Gemma-4-26B-A4B-NVFP4NVFP4vllm17.5 GiB on diskquantized by nvidia
hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q8_0Q8_0ollama25.0 GiB on diskquantized by unsloth
RedHatAI/gemma-4-26B-A4B-it-FP8-dynamicFP8vllm26.7 GiB on diskquantized by RedHatAI
unsloth/gemma-4-26B-A4B-it-GGUF:BF16BF16llamacpp47.0 GiB on diskquantized by unsloth
hf.co/ggml-org/gemma-4-26B-A4B-it-GGUF:BF16BF16ollama47.0 GiB on diskquantized by ggml-org
Gemma 4 31B
hf.co/unsloth/gemma-4-31B-it-GGUF:IQ2_XXSIQ2_XXSollama7.9 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-31B-it-GGUF:IQ2_MIQ2_Mollama10.0 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-31B-it-GGUF:IQ3_XXSIQ3_XXSollama11.0 GiB on diskquantized by unsloth
hf.co/google/gemma-4-31B-it-qat-q4_0-gguf:Q4_0Q4_0ollama16.4 GiB on diskpublished by google
hf.co/unsloth/gemma-4-31B-it-GGUF:Q4_K_MQ4_K_Mollama17.1 GiB on diskquantized by unsloth
google/gemma-4-31B-it-qat-w4a16-ctW4A16vllm21.7 GiB on diskpublished by google
nvidia/Gemma-4-31B-IT-NVFP4NVFP4vllm30.4 GiB on diskquantized by nvidia
hf.co/unsloth/gemma-4-31B-it-GGUF:Q8_0Q8_0ollama30.4 GiB on diskquantized by unsloth
unsloth/gemma-4-31B-it-GGUF:BF16BF16llamacpp57.2 GiB on diskquantized by unsloth
google/gemma-4-31B-it-qat-q4_0-unquantizedBF16vllm58.3 GiB on diskpublished by google
Gemma 4 ASSISTANT-12B
google/gemma-4-12B-it-assistantBF16vllm806 MiB on diskpublished by google
Gemma 4 ASSISTANT-26B-A4B
google/gemma-4-26B-A4B-it-assistantBF16vllm800 MiB on diskpublished by google
Gemma 4 E2B5.1B
hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ2_MIQ2_Mollama2.1 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ3_XXSIQ3_XXSollama2.2 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XSIQ4_XSollama2.8 GiB on diskquantized by unsloth
mlx-community/gemma-4-E2B-it-4bit4bitmlx2.8 GiB on diskquantized by mlx-community
hf.co/google/gemma-4-E2B-it-qat-q4_0-gguf:Q4_0Q4_0ollama3.1 GiB on diskpublished by google
hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_MQ4_K_Mollama3.8 GiB on diskquantized by unsloth
lmstudio-community/gemma-4-E2B-it-MLX-4bit4BITmlx4.0 GiB on diskquantized by lmstudio-community
lmstudio-community/gemma-4-E2B-it-MLX-5bit5BITmlx4.4 GiB on diskquantized by lmstudio-community
hf.co/unsloth/gemma-4-E2B-it-GGUF:Q8_0Q8_0ollama4.7 GiB on diskquantized by unsloth
lmstudio-community/gemma-4-E2B-it-MLX-8bit8BITmlx5.5 GiB on diskquantized by lmstudio-community
gemma4:e2bQ4_K_Mollama7.0 GiB on diskpublished by google
google/gemma-4-E2B-it-qat-w4a16-ctW4A16vllm7.8 GiB on diskpublished by google
hf.co/unsloth/gemma-4-E2B-it-GGUF:BF16BF16ollama8.7 GiB on diskquantized by unsloth
gemma-4-E2B-it-Artemis-GGUF
hf.co/turintech/gemma-4-E2B-it-Artemis-GGUF:IQ2_MIQ2_Mollama1.7 GiB on diskderived by turintech
hf.co/turintech/gemma-4-E2B-it-Artemis-GGUF:IQ3_XXSIQ3_XXSollama1.9 GiB on diskderived by turintech
hf.co/turintech/gemma-4-E2B-it-Artemis-GGUF:Q3_K_LQ3_K_Lollama2.1 GiB on diskderived by turintech
Gemma 4 E4B8B
hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ2_MIQ2_Mollama3.3 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ3_XXSIQ3_XXSollama3.5 GiB on diskquantized by unsloth
hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ4_XSIQ4_XSollama4.4 GiB on diskquantized by unsloth
mlx-community/gemma-4-E4B-it-4bit4bitmlx4.5 GiB on diskquantized by mlx-community
hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_K_MQ4_K_Mollama4.6 GiB on diskquantized by unsloth
hf.co/google/gemma-4-E4B-it-qat-q4_0-gguf:Q4_0Q4_0ollama4.8 GiB on diskpublished by google
lmstudio-community/gemma-4-E4B-it-MLX-4bit4BITmlx6.4 GiB on diskquantized by lmstudio-community
lmstudio-community/gemma-4-E4B-it-MLX-6bit6BITmlx7.3 GiB on diskquantized by lmstudio-community
hf.co/unsloth/gemma-4-E4B-it-GGUF:Q8_0Q8_0ollama7.6 GiB on diskquantized by unsloth
lmstudio-community/gemma-4-E4B-it-MLX-8bit8BITmlx8.3 GiB on diskquantized by lmstudio-community
google/gemma-4-E4B-it-qat-w4a16-ctW4A16vllm10.7 GiB on diskpublished by google
gemma4:e4bQ4_K_Mollama10.7 GiB on diskpublished by google
hf.co/unsloth/gemma-4-E4B-it-GGUF:BF16BF16ollama14.0 GiB on diskquantized by unsloth
gemma-4-E4B-it-Artemis-GGUF
hf.co/turintech/gemma-4-E4B-it-Artemis-GGUF:Q3_K_MQ3_K_Mollama3.3 GiB on diskderived by turintech
hf.co/turintech/gemma-4-E4B-it-Artemis-GGUF:Q4_0Q4_0ollama3.9 GiB on diskderived by turintech
ask your own gram

Your concierge can answer this from the network rather than from this page. The ask carries the operator id, so it resolves to a capability and an engine — not to a name somebody has to recognise.

What has this network measured about the Gemma 4 family (epn.family.gemma-4), and can my gram run any of it? See /inferences/gemma-4

open a gramx with this ask →epn.family.gemma-4