Topic
Reliance on provider-published scores and unpublished internal testing in place of independent evaluation of coding models.
Ternary weights at 1.76 bits put a 27B model into a 5.9GB file and let a laptop decode it at 28.1 tokens a second. The retention figure comes from Prism ML's own benchmark suite, not the table on the model card.
Reality
Perplexity pairs about 190 million web pages with 69,721 agent-written queries, which is closer to production than most retrieval tests get, and keeps the corpus, queries and labels private so it stays the only party able to run it.
Reality
The vLLM benchmark shows a finished MP4 arriving before its own playback would end, which is a different property from showing frames as they are made, and MiniMax's community licence still requires separate permission for US, EU, UK and Korean use.
Reality
The release attaches speculative decoding to weights it publishes under MIT, which lowers what a self-hosting threat costs to stand up, even though the performance claim behind it is still the vendor's own.
Reality
Anthropic held the price on Claude Opus 4.8, cut fast mode to a third of its previous cost, and handed users an effort dial. That combination is what gets agents into engineering budgets.
Reality
A dev.to post sells numpy_cache as 20x faster than np.savez_compressed and half the size of np.save. Its own baselines put the write nearer 1.5 ms per megabyte, and the file format lets you check the rest.
Reality
Intern-S2-Preview-397B puts page-level paper reading and tool use into Apache 2.0 weights. The download listing says 810 GB, and every benchmark number so far is the lab's own.
Reality
Etched has first-pass silicon, 400 staff and more than $1bn in booked orders. What it does not have in public is a peak FLOPS figure, a power draw, or a third-party benchmark.
Reality
A read-only sensor for cross-tenant memory leaks ships with no tagged release, because its own backend cannot enumerate what the retriever exposed. That limitation is the story.
Reality
GitHub added xAI's model to Copilot on August 14 across eight developer surfaces at usage-based pricing. The benchmark case, including xAI's own terminal scores, is mixed.
Reality