Every 7 days the grams on this network work the model catalog in cohorts — each family's lead member at rung 0 on every engine first, then its quantization ladder, then the other members, then derivatives — and publish what they proved, per hardware class, with how many machines contributed. No number here is from a model card. Publishers are listed by what the network proved of their builds, never by popularity.
2026-W44
Open now. Grams take the least-covered artifact for their own hardware class first, so two machines of one kind never measure the same thing while a third kind waits.
audio, audio.in, audio.out, lang.en, lang.hi, lang.mr, lang.ne, lang.ta, lang.te, lang.ur
code.reason, code.write, embed.vector, faithful_tool_result, instruct.follow, lang.bn, lang.en, lang.gu, lang.hi, lang.kn, lang.ml, lang.mr, lang.nl, lang.pa …
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
vision.count, vision.spatial
image.generate
voice.identify
audio, audio.in, lang.en
audio, audio.out, lang.en
per engine: how many artifacts this season proved, on how many hardware classes, which capabilities came back, and how many artifact-slots the fleet currently cannot reach at all (no Apple Silicon for MLX, no schedulable card for vLLM, no cluster for the speech pods). An engine with zero proved and a large unreachable count has never been shown to work on this network -- a statement about the network, not about the models.
away 43 h
expects 221 h for the rest (own samples) · trailing rate says 1388 h · blocked: fetch
expects 410 h for the rest (own samples) · trailing rate says 3188 h · blocked: fetch
away 56 h · last session ended without being asked to — power loss, a hard reset, or a crash
expects 23 h for the rest (own samples) · trailing rate says 648 h
expects 280 h for the rest (own samples) · trailing rate says 2620 h · blocked: engine
away 257 h · last session a supervisor bounce: a new session started while the last one was alive · this machine sleeps (46 min during its last session) — most likely asleep
expects 23 h for the rest (own samples) · blocked: budget
away 447 h · last session ended without being asked to — power loss, a hard reset, or a crash · this machine sleeps (37 min during its last session) and is on battery — most likely asleep
expects 53 h for the rest (own samples) · trailing rate says 1500 h · blocked: fetch
away 4 h · last session a supervisor bounce: a new session started while the last one was alive
expects 18 h for the rest (own samples) · blocked: budget
expects 1079 h for the rest (own samples) · blocked: budget
away 183 h · last session ended without being asked to — power loss, a hard reset, or a crash
no priced estimate yet · trailing rate says 3336 h · blocked: pace
away 128 h · last session ended without being asked to — power loss, a hard reset, or a crash
expects 112 h for the rest (own samples) · blocked: pace
away 16 h · last session a supervisor bounce: a new session started while the last one was alive · a laptop on battery — most likely asleep
no priced estimate yet · blocked: pace
away 83 h · last session a supervisor bounce: a new session started while the last one was alive · a laptop on battery — most likely asleep
expects 7 h for the rest (own samples) · blocked: pace
away 134 h · last session a supervisor bounce: a new session started while the last one was alive · this machine sleeps (10 min during its last session) and is on battery — most likely asleep
expects 113 h for the rest (own samples) · blocked: pace
away 44 h · last session a supervisor bounce: a new session started while the last one was alive
expects 42 h for the rest (own samples) · blocked: pace
one row per gram: what it is, what its hardware can never reach, how far it has got, and its OWN priced estimate for the rest. There is no single fleet ETA, because a gram only proves what it can obtain.
proving is a LEASE, not a fact: what each gram said it was working on when it last reported. It expires on its own, confers nothing and is never coverage. It is here so the network can be watched working.
64
cells signed as gated, unloadable, or not published upstream. Each one is a measurement every other machine of that class is spared.
Medians from the grams own clocks, carried across windows so a class whose machines are asleep still plans on what the network paid to learn.
Tokens actually served, cumulative. Demand decides how deep a proof goes and where weights sit — never what exists.
Past seasons
The season files are public JSON, built by an open aggregator from signed results anyone can re-verify — data/catalog/seasons on GitHub. Cite them; they carry the signatures. Back to the evidence table →