2 Nov 2026 to 9 Nov 2026. What the grams proved of the catalog this window, per artifact and hardware class. Lineage is from the catalog; every number is a signed measurement or a count of them, and a gram count is a verified floor.
The order a season proves things in: the number every other number is compared to comes first.
A class is os · arch · card class · RAM band. A number proved on one says nothing about another.
Listed by held outcomes. Downloads, likes and popularity do not appear on this page, because nobody signs them.
the engine has no loader for this architecture — a release changes that, never a clock
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the ref, or the quant asked of it, is not published upstream
the engine has no loader for this architecture — a release changes that, never a clock
the engine has no loader for this architecture — a release changes that, never a clock
the engine has no loader for this architecture — a release changes that, never a clock
the engine has no loader for this architecture — a release changes that, never a clock
the engine has no loader for this architecture — a release changes that, never a clock
the engine has no loader for this architecture — a release changes that, never a clock
the engine has no loader for this architecture — a release changes that, never a clock
the ref, or the quant asked of it, is not published upstream
the engine has no loader for this architecture — a release changes that, never a clock
the engine has no loader for this architecture — a release changes that, never a clock
the engine has no loader for this architecture — a release changes that, never a clock
refusals are signed facts about (artifact, class): gated, unloadable, or absent upstream. They are coverage of a kind -- the fleet learned it once -- and they sink the cell to the back of every gram of that class. A refusal lifts when its season closes and the engine that refused has moved on.
carried forward — no gram of this class reported in the latest aggregate
carried forward — no gram of this class reported in the latest aggregate
carried forward — no gram of this class reported in the latest aggregate
cost is operational: medians of node-reported pull/probe clocks per hardware class and engine, with sample counts. It is not a measurement of any model and is never ranked.
demand is tokens produced per artifact per class, from signed throughput observations; it orders depth, never existence.
Cohort 0 · Lead member, rung 0 · 53
axteam · quantization of zai-org/GLM-OCR
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, lang.en, output.long, output.schema, structured_output, vision, vision.count, vision.multi_image, vision.read, vision.spatial
coqui · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: audio, audio.out
refused on Windows · x86-64 · NVIDIA · 16 GiB — unavailable, by 1 gram
refused on macOS · ARM64 · Apple GPU · 16 GiB — unavailable, by 1 gram
refused on Windows · x86-64 · AMD · 8 GiB — unavailable, by 1 gram
refused on Linux · x86-64 · no GPU · 32 GiB — unloadable, by 1 gram
depth-anything · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: vision.spatial
Efficient-Large-Model · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: image.generate
facebook · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: audio, audio.out
facebook · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: translate.as, translate.bn, translate.gu, translate.hi, translate.kn, translate.ml, translate.mr, translate.or, translate.pa, translate.ta, translate.te, translate.text, translate.ur
facebook · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: vision.count
facebook · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: audio, audio.in, lang.as, lang.bn, lang.en, lang.gu, lang.kn, lang.ml, lang.mr, lang.ne, lang.or, lang.pa, lang.ta, lang.te, lang.ur
google · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated, vision, vision.multi_image, vision.read, vision.spatial
served 282.3k tokens across the network
google · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: translate.as, translate.bn, translate.gu, translate.hi, translate.kn, translate.ml, translate.mr, translate.or, translate.pa, translate.ta, translate.text
bartowski · quantization of HuggingFaceTB/SmolLM3-3B
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, instruct.follow, lang.en, lang.hi, output.long, output.schema, research.synthesize, structured_output
served 29.0k tokens across the network
bartowski · quantization of krutrim-ai-labs/Krutrim-2-instruct
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, research.synthesize, structured_output
served 3.2k tokens across the network
bartowski · quantization of sarvamai/sarvam-1
held on Linux · x86-64 · NVIDIA · 64 GiB: lang.bn, lang.gu, lang.hi, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, structured_output
served 12.5k tokens across the network
bullerwins · quantization of google/translategemma-4b-it
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.mr, lang.pa, output.long, output.schema, structured_output
served 2.9k tokens across the network
ibm-granite · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.en, lang.kn, lang.ta, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated
served 22.0k tokens across the network
meta-models · vendor build
served 174 tokens across the network
mistralai · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, instruct.follow, lang.en, lang.te, output.long, tools, tools.call, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 3.9k tokens across the network
mradermacher · quantization of ai4bharat/Airavata
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, instruct.follow, lang.en, lang.hi, output.long
served 26.3k tokens across the network
mradermacher · quantization of zai-org/GLM-OCR
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, lang.en, output.schema, structured_output, vision, vision.multi_image, vision.read
served 10.4k tokens across the network
ngquocvinh · quantization of google/embeddinggemma-2
held on Linux · x86-64 · NVIDIA · 64 GiB: embed.vector
openbmb · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, instruct.follow, lang.bn, lang.en, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call
served 9.0k tokens across the network
openbmb · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, instruct.follow, lang.en, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.loop_ends, tools_loop_terminated, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 19.2k tokens across the network
openbmb · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, instruct.follow, lang.en, output.long, output.schema, structured_output, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 14.7k tokens across the network
openbmb · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, instruct.follow, lang.en, output.long, output.schema, reason.channel, structured_output, thinking, tools, tools.call, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 37.9k tokens across the network
openbmb · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, instruct.follow, lang.bn, lang.en, output.long, output.schema, structured_output, tools, tools.call, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 2.3k tokens across the network
openbmb · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, lang.en, lang.hi, lang.ml, lang.ta, lang.te, output.long, reason.channel, research.synthesize, thinking
served 19.2k tokens across the network
openbmb · vendor build
held on Windows · x86-64 · NVIDIA · 16 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.en, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 68.2k tokens across the network
QuantFactory · quantization of sarvamai/OpenHathi-7B-Hi-v0.1-Base
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason
served 155.7k tokens across the network
Qwen · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, faithful_tool_result, instruct.follow, lang.en, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 237.5k tokens across the network
Qwen · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.en, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.multi_hop
served 46.4k tokens across the network
Reza2kn · quantization of baidu/Qianfan-OCR
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, instruct.follow, lang.en, lang.hi, lang.kn, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 3.9k tokens across the network
sabafallah · quantization of deepseek-ai/DeepSeek-OCR-2
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, lang.en, output.long, output.schema, structured_output
served 40.1k tokens across the network
sabafallah · quantization of deepseek-ai/DeepSeek-OCR
held on Linux · x86-64 · NVIDIA · 64 GiB: lang.en, output.schema, research.synthesize, structured_output
served 18.2k tokens across the network
second-state · quantization of google/embeddinggemma-300m
held on Linux · x86-64 · NVIDIA · 64 GiB: embed.vector
tencent · vendor build
served 1.4k tokens across the network
tencent · vendor build
served 31.9k tokens across the network
unsloth · quantization of google/functiongemma-270m-it
held on Linux · x86-64 · NVIDIA · 64 GiB: lang.en, lang.hi, output.schema, structured_output
served 34.7k tokens across the network
unsloth · quantization of google/gemma-3-270m-it
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, lang.en, output.schema, structured_output
served 15.5k tokens across the network
unsloth · quantization of google/gemma-4-E2B-it
held on Linux · x86-64 · NVIDIA · 256 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 512.1k tokens across the network
unsloth · quantization of zai-org/GLM-4.7-Flash
held on Linux · x86-64 · NVIDIA · 32 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.hi, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated
served 84.3k tokens across the network
unsloth · quantization of openai/gpt-oss-20b
held on Linux · x86-64 · NVIDIA · 32 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated
served 47.7k tokens across the network
unsloth · quantization of ibm-granite/granite-4.1-3b
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, faithful_tool_result, instruct.follow, lang.en, lang.kn, lang.te, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.multi_hop
served 2.2k tokens across the network
unsloth · quantization of meta-llama/Llama-3.2-3B-Instruct
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.hi, lang.ml, lang.pa, lang.ta, lang.te, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 4.0k tokens across the network
unsloth · quantization of mistralai/Magistral-Small-2506
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, research.synthesize, structured_output, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 31.7k tokens across the network
unsloth · quantization of google/medgemma-1.5-4b-it
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, research.synthesize, structured_output, vision, vision.count, vision.multi_image, vision.read
served 6.6k tokens across the network
unsloth · quantization of microsoft/Phi-4-mini-instruct
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, instruct.follow, lang.en, lang.hi, output.long, output.schema, structured_output
served 17.6k tokens across the network
unsloth · quantization of Qwen/Qwen3.5-4B
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, instruct.follow, lang.bn, lang.en, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, structured_output, thinking, tools, tools.call, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 18.3k tokens across the network
unsloth · quantization of Qwen/Qwen3.6-27B
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated
served 17.0k tokens across the network
pyannote · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: voice.identify
speechbrain · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: voice.identify
stabilityai · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: image.generate
refused on Windows · x86-64 · Intel GPU · 16 GiB — unloadable, by 1 gram
SWivid · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: audio, audio.out, lang.en
refused on Windows · x86-64 · Intel GPU · 16 GiB — unloadable, by 1 gram
refused on Windows · x86-64 · NVIDIA · 16 GiB — unavailable, by 1 gram
refused on macOS · ARM64 · Apple GPU · 16 GiB — unavailable, by 1 gram
refused on Linux · x86-64 · no GPU · 32 GiB — unloadable, by 1 gram
THUDM · vendor build
refused on Linux · x86-64 · NVIDIA · 32 GiB — unloadable, by 1 gram
Cohort 1 · Lead member's ladder · 13
AlphaOxO · quantization of zai-org/GLM-OCR
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, lang.en, output.schema, structured_output, vision, vision.read
facebook · vendor build
openbmb · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: code.write, instruct.follow, lang.en, lang.ta, output.long, output.schema, structured_output, tools, tools.call, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 7.0k tokens across the network
unsloth · quantization of openai/gpt-oss-20b
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.ha, lang.hi, lang.kn, lang.nl, lang.ru, lang.sw, lang.ta, lang.te, output.schema, reason.channel, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 14.0k tokens across the network
unsloth · quantization of Qwen/Qwen3.8-27B
held on Linux · x86-64 · NVIDIA · 32 GiB: code.reason, code.write, instruct.follow, lang.am, lang.bn, lang.en, lang.ha, lang.hi, lang.kn, lang.mr, lang.nl, lang.ru, lang.sw, lang.ta, lang.te, lang.yo, output.schema, reason.channel, structured_output, thinking, tools, tools.call
served 4.4k tokens across the network
unsloth · quantization of openai/gpt-oss-20b
held on Linux · x86-64 · NVIDIA · 32 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.hi, lang.kn, lang.ml, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated
served 45.3k tokens across the network
unsloth · quantization of openai/gpt-oss-20b
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.as, lang.bn, lang.en, lang.es, lang.hi, lang.ja, lang.kn, lang.mr, lang.ne, lang.nl, lang.ru, lang.ta, lang.te, lang.ur, lang.zh, output.schema, reason.channel, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 33.1k tokens across the network
unsloth · quantization of openai/gpt-oss-20b
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.hi, lang.kn, lang.nl, lang.ru, lang.ta, lang.te, output.schema, reason.channel, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 32.6k tokens across the network
unsloth · quantization of google/embeddinggemma-300m
held on Linux · x86-64 · NVIDIA · 256 GiB: embed.vector
unsloth · quantization of google/gemma-4-E2B-it
held on Windows · x86-64 · NVIDIA · 16 GiB: faithful_tool_result, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 37.1k tokens across the network
unsloth · quantization of openai/gpt-oss-20b
held on Linux · x86-64 · NVIDIA · 32 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated
served 15.2k tokens across the network
kokoro · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: audio, audio.out, lang.en
UsefulSensors · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: audio, audio.in, lang.en
Cohort 2 · Other vendor members · 12
depth-anything · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: vision.spatial
facebook · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: translate.as, translate.bn, translate.gu, translate.hi, translate.kn, translate.ml, translate.mr, translate.or, translate.pa, translate.ta, translate.te, translate.text, translate.ur
ibm-granite · vendor build
held on Linux · x86-64 · NVIDIA · 32 GiB: output.schema, structured_output
served 1.3k tokens across the network
Qwen · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: faithful_tool_result, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 3.4k tokens across the network
unsloth · quantization of Qwen/Qwen3.5-35B-A3B
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, instruct.follow, lang.bn, lang.en, lang.es, lang.ha, lang.hi, lang.kn, lang.mr, lang.ne, lang.nl, lang.si, lang.ta, lang.te, lang.ur, reason.channel, thinking, tools, tools.call
served 18.2k tokens across the network
openai · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: audio, audio.in, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.mr, lang.ne, lang.pa, lang.ta, lang.ur
openai · vendor build
held on macOS · ARM64 · Apple GPU · 16 GiB: audio, audio.in, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.mr, lang.ne, lang.ta, lang.te, lang.ur
openai · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: audio, audio.in, lang.en, lang.hi, lang.mr, lang.ta
openai · vendor build
held on Linux · x86-64 · NVIDIA · 64 GiB: audio, audio.in, lang.en, lang.hi, lang.mr, lang.ne, lang.ta, lang.ur
facebook · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: audio.music
unsloth · quantization of Qwen/Qwen3.5-35B-A3B
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, instruct.follow, lang.bn, lang.en, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, structured_output, thinking, tools, tools.call
served 16.0k tokens across the network
UsefulSensors · vendor build
held on Linux · x86-64 · no GPU · 32 GiB: audio, audio.in, lang.en
Cohort 3 · Derivatives · 14
Beh1222 · derivative of a vendor base
held on Linux · x86-64 · NVIDIA · 64 GiB: image.generate
llmfan46 · derivative of llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, vision, vision.count, vision.read, vision.spatial
served 13.3k tokens across the network
sarvamai · derivative of google/gemma-3-4b-it
held on Linux · x86-64 · NVIDIA · 64 GiB: translate.as, translate.bn, translate.gu, translate.hi, translate.kn, translate.ml, translate.mr, translate.ne, translate.or, translate.pa, translate.ta, translate.te, translate.text, translate.ur
SPRINGLab · derivative of SWivid/F5-TTS
held on Linux · x86-64 · NVIDIA · 64 GiB: audio, audio.out
segmind · derivative of a vendor base
held on Linux · x86-64 · no GPU · 32 GiB: image.generate
llmfan46 · derivative of llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, instruct.follow, lang.bn, lang.en, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, vision, vision.count, vision.read, vision.spatial
served 14.7k tokens across the network
mradermacher · derivative of CohereLabs/North-Mini-Code-1.0
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call
served 5.1k tokens across the network
TeichAI · derivative of TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill
held on Linux · x86-64 · no GPU · 32 GiB: code.write, instruct.follow, lang.bn, lang.en, lang.hi, lang.kn, lang.ta, lang.te, output.long, output.schema, research.synthesize, structured_output
served 777 tokens across the network
huihui-ai · derivative of Qwen/Qwen3.6-35B-A3B
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, lang.am, lang.as, lang.bn, lang.en, lang.es, lang.gu, lang.hi, lang.kn, lang.mr, lang.ne, lang.nl, lang.ru, lang.si, lang.ta, lang.te, lang.ur, lang.yo, output.schema, reason.channel, structured_output, thinking, tools, tools.call, vision, vision.count, vision.read, vision.spatial
served 31.7k tokens across the network
llmfan46 · derivative of llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 13.4k tokens across the network
mradermacher · derivative of tencent/HY-MT1.5-1.8B
served 3.3k tokens across the network
nm-testing · derivative of a vendor base
yujiepan · derivative of tencent/Hy3
huihui-ai · derivative of Qwen/Qwen3.6-35B-A3B
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, instruct.follow, lang.bn, lang.en, lang.ta, output.long, reason.channel, research.synthesize, thinking, tools, tools.call, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 17.8k tokens across the network
Measured, not in the catalog · 16
unknown publisher
held on hardware not stated in the signed payload: tools
served 359.2k tokens across the network
unknown publisher
held on hardware not stated in the signed payload: tools
served 36.4k tokens across the network
unknown publisher
held on hardware not stated in the signed payload: tools
served 48.8k tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.hi, lang.te, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 5.0k tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 GiB: faithful_tool_result, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 397 tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, instruct.follow, lang.bn, lang.en, lang.hi, lang.nl, lang.ru, output.schema, structured_output
served 9.7k tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, faithful_tool_result, instruct.follow, lang.en, lang.nl, lang.ru, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 223.2k tokens across the network
unknown publisher
held on Linux · x86-64 · NVIDIA · 256 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated, vision, vision.count, vision.multi_image, vision.read, vision.spatial
served 327.8k tokens across the network
unknown publisher
held on Linux · x86-64 · no GPU · 32 GiB: code.write, faithful_tool_result, instruct.follow, lang.en, lang.hi, lang.ta, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 109.7k tokens across the network
unknown publisher
held on Windows · x86-64 · Intel GPU · 16 GiB: audio, audio.out, lang.en
unknown publisher
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, lang.bn, lang.en, lang.hi, lang.kn, lang.ta, lang.te, output.long, output.schema, reason.channel, structured_output, thinking, tools, tools.call
unknown publisher
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, lang.bn, lang.en, lang.hi, lang.kn, lang.mr, lang.ta, lang.te, output.long, output.schema, reason.channel, structured_output, thinking, tools, tools.call
unknown publisher
held on Linux · x86-64 · no GPU · 32 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.multi_hop
unknown publisher
held on Linux · x86-64 · NVIDIA · 64 GiB: code.reason, code.write, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.pa, lang.ta, lang.te, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call
unknown publisher
held on Linux · x86-64 · no GPU · 32 GiB: image.generate
unknown publisher
held on Linux · x86-64 · NVIDIA · 64 GiB: image.generate
Every number is a signed measurement or a count of them. grams is how many distinct nodes contributed: a verified floor, not a complete account. Lineage (publisher, attribution, base, cohort, rung) is from the catalog and is not a measurement.
grams and classes are what THIS window heard from; coverage, refusals, cost and demand carry across windows, because a gram going to sleep does not unmake what it measured.