14 Sept 2026 to 21 Sept 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 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
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 · 10
depth-anything · vendor build
Efficient-Large-Model · vendor build
facebook · vendor build
google · vendor build
held on Windows · x86-64 · NVIDIA · 16 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 213.2k tokens across the network
bartowski · quantization of sarvamai/sarvam-1
held on Windows · x86-64 · NVIDIA · 16 GiB: instruct.follow, lang.bn, lang.hi, lang.kn, lang.mr, lang.te, output.long, output.schema, structured_output
served 425 tokens across the network
MaziyarPanahi · quantization of microsoft/Phi-4-mini-instruct
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, instruct.follow, lang.bn, lang.en, output.long, output.schema, research.synthesize, structured_output
served 66 tokens across the network
Qwen · vendor build
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 18.5k tokens across the network
unsloth · quantization of google/gemma-4-E2B-it
held on Windows · x86-64 · NVIDIA · 16 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.am, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.nl, lang.pa, lang.ru, lang.sw, 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 58.7k tokens across the network
THUDM · vendor build
Wan-AI · vendor build
Cohort 1 · Lead member's ladder · 14
facebook · vendor build
held on macOS · ARM64 · Apple GPU · 16 GiB: audio.music
MaziyarPanahi · quantization of microsoft/Phi-4-mini-instruct
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, instruct.follow, lang.bn, lang.en, output.long, output.schema, structured_output
served 22 tokens across the network
Qwen · vendor build
held on Windows · x86-64 · NVIDIA · 16 GiB: faithful_tool_result, output.schema, reason.channel, structured_output, thinking, tools, tools.call, tools.faithful, tools.multi_hop
served 2.6k tokens across the network
MaziyarPanahi · quantization of microsoft/Phi-4-mini-instruct
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, instruct.follow, lang.en, output.long, output.schema, research.synthesize, structured_output
served 163 tokens across the network
unsloth · quantization of Qwen/Qwen3-4B
held on Windows · x86-64 · NVIDIA · 16 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.en, lang.ta, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 124.5k tokens across the network
MaziyarPanahi · quantization of microsoft/Phi-4-mini-instruct
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, instruct.follow, lang.bn, lang.en, output.long, output.schema, research.synthesize, structured_output
served 27 tokens across the network
Qwen · vendor build
held on Windows · x86-64 · NVIDIA · 16 GiB: faithful_tool_result, output.schema, reason.channel, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools.multi_hop, tools_loop_terminated
served 1.4k tokens across the network
unsloth · quantization of microsoft/Phi-4-mini-instruct
held on Windows · x86-64 · NVIDIA · 16 GiB: output.schema, structured_output
served 22 tokens across the network
unsloth · quantization of Qwen/Qwen3-4B
held on Windows · x86-64 · NVIDIA · 16 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.en, lang.ta, output.long, output.schema, reason.channel, research.synthesize, structured_output, thinking, tools, tools.call, tools.faithful, tools.multi_hop
served 2.4k tokens across the network
MaziyarPanahi · quantization of microsoft/Phi-4-mini-instruct
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, lang.en, output.long, output.schema, structured_output
served 36 tokens across the network
unsloth · quantization of Qwen/Qwen3-4B
served 70.2k tokens across the network
MaziyarPanahi · quantization of microsoft/Phi-4-mini-instruct
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, instruct.follow, lang.en, output.long, output.schema, structured_output
served 59 tokens across the network
Qwen · vendor build
held on Windows · x86-64 · NVIDIA · 16 GiB: faithful_tool_result, output.schema, reason.channel, structured_output, thinking, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 23.8k tokens across the network
kokoro · vendor build
held on Windows · x86-64 · Intel GPU · 16 GiB: audio, audio.out
Cohort 2 · Other vendor members · 5
lmstudio-community · quantization of microsoft/Phi-4-mini-reasoning
held on Windows · x86-64 · AMD · 8 GiB: code.reason, output.schema, reason.channel, research.synthesize, structured_output, thinking
served 13.5k tokens across the network
stabilityai · vendor build
unsloth · quantization of microsoft/Phi-4-mini-reasoning
held on Windows · x86-64 · NVIDIA · 16 GiB: lang.en, lang.es, output.schema, reason.channel, structured_output, thinking
served 5.3k tokens across the network
unsloth · quantization of microsoft/Phi-4-mini-reasoning
served 35.0k tokens across the network
lmstudio-community · quantization of microsoft/Phi-4-mini-reasoning
held on Windows · x86-64 · AMD · 8 GiB: output.schema, structured_output
served 6.3k tokens across the network
Cohort 3 · Derivatives · 1
segmind · derivative of a vendor base
Measured, not in the catalog · 16
unknown publisher
held on hardware not stated in the signed payload: tools
served 100.8k tokens across the network
unknown publisher
held on hardware not stated in the signed payload: tools
served 5.3k tokens across the network
unknown publisher
held on hardware not stated in the signed payload: tools
unknown publisher
unknown publisher
held on macOS · ARM64 · Apple GPU · 16 GiB: code.reason, instruct.follow, lang.en, lang.hi, lang.kn, output.long, output.schema, research.synthesize, structured_output
served 3.9k tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, faithful_tool_result, instruct.follow, lang.en, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 42.0k tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 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 786 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.mr, lang.te, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 419 tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 GiB: lang.en
served 191 tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 GiB: code.reason, instruct.follow, lang.en, output.long, output.schema, structured_output
served 24 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.mr, lang.pa, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 151 tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 GiB: code.write, faithful_tool_result, lang.en, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 174 tokens across the network
unknown publisher
held on Windows · x86-64 · NVIDIA · 16 GiB: code.reason, code.write, faithful_tool_result, instruct.follow, lang.bn, lang.en, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 189 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.te, output.long, output.schema, research.synthesize, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 165 tokens across the network
unknown publisher
held on Windows · x86-64 · AMD · 8 GiB: code.write, faithful_tool_result, instruct.follow, lang.en, lang.hi, output.long, output.schema, structured_output, tools, tools.call, tools.faithful, tools.loop_ends, tools_loop_terminated
served 40.4k tokens across the network
unknown publisher
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.