One signal

Update What changed

The Decoder says Moonshot AI’s Kimi K3.1 Pro scored 55 on the Artificial Analysis Intelligence Index, an 11-point increase over its predecessor. That gives the reported fourth-place ranking a useful measure of the jump, rather than just a spot on the table.

The article says only three models from Anthropic and Meta rank higher. Those figures are reported benchmark results, not an independent assessment of how Kimi performs across the tasks readers might care about.

The Decoder also says that, if Moonshot releases the weights as planned, the gap between open and proprietary models would be three points. That makes the forthcoming weights release worth watching: it could let developers inspect and run the model, while the leaderboard alone cannot settle how useful it will be in practice.

Sources and evidence

• Chinese AI model Kimi K3.1 Pro is closing in on Anthropic's Claude Opus 5.5: The Decoder reports that Kimi K3.1 Pro scored 55 on the Artificial Analysis Intelligence Index, 11 points above its predecessor, and says a planned weights release could leave it three points behind proprietary models.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

R&D World reports that Kimi K3.1 Pro costs $2 per million input tokens and $10 per million output tokens, down from $3 and $15 for Kimi K3. The report says those rates match OpenAI’s GPT-6.1 Sol and are half the price of Claude Opus 5.5.

Artificial Analysis measured K3.1 Pro at 77 tokens per second and calculated a weighted average of $2.11 per Intelligence Index task. R&D World compares that with $5.98 for Opus 5.5, $7.63 for Claude Fable 5.1 and $0.72 for GPT-6.1 Sol.

The report also describes a mixed benchmark picture: Moonshot’s own sheet gives K3.1 Pro leading scores on six of 14 tests, while Arena AI capability lead Peter Gostev said its results in his tests fell below GPT-6.1 Sol. R&D World notes that the comparison sheet combines vendor and public results, and that some rival scores may involve fallbacks or safeguards.

Sources and evidence

• Kimi K3.1 Pro vs. Claude Opus 5.5 and GPT-6 Astra - R&D World: R&D World reports new pricing and benchmark details for Moonshot’s Kimi K3.1 Pro, including a $2.11 weighted average cost per Artificial Analysis Intelligence Index task and a mixed comparison across benchmark sources.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

Moonshot AI says its Kimi K3.1 Pro model improved a simulated human’s walking by finding and correcting 36 errors in the simulator’s shoulder-joint calculations. In the company’s account, the simulated figure eventually managed alternating strides for 11.2 seconds.

The model was working with human-atlas, an open-source anatomical model, after attempts to improve arm motion through changes to its training rewards failed. Moonshot says K3.1 Pro then investigated the simulator itself and retrained the walking controller after correcting the errors.

It is an intriguing example of a language model doing more than adjusting training settings, but the result is a company-described internal simulation, not evidence of a robot learning to walk in the real world.

Sources and evidence

• Kimi K3.1 Pro vs. Claude Opus 5.5 and GPT-6 Astra - R&D World: Moonshot AI’s technical account says Kimi K3.1 Pro found and corrected 36 errors in calculations governing simulated shoulder-joint rotation, then retrained a walking controller; the simulated human achieved alternating strides for 11.2 seconds.

Independent WittyWires Watcher; not an official account or feed.