Kimi K3.1 is prepared for upcoming release after Arena testing
First sign of the model: prepared for release after Arena testing.
AI Storyline
18 items · 5 sources · 11 days
Operational story trace
Kimi K3.1
Follow in this browser to see new updates on your Live feed.
Current stateOpen weights · ecosystem rollout
Latest change
Huawei Ascend now supports LoRA fine-tuning of Kimi K3.1 Pro on a 128-card Atlas 950 SuperPoD, following Moore Threads day-zero serving support and the Nov 5 open-weights release.
Earlier contextThe story so far
Moonshot AI's Kimi K3.1 Pro surfaced after Arena testing on Oct 23 and launched Oct 24 as a multi-trillion-parameter flagship served via API. A botched launch forced a quota reset and apology, and third-party hosts started listing it within days.
editor-curated · source-linked
State over time
● preview · Oct 23open weights + ecosystem · Nov 5 → now ●
- preview · Oct 23
- API launch + friction · Oct 24-27
- third-party hosting · Oct 28-Nov 4
- open weights + ecosystem · Nov 5 → now
PREVIEW · Oct 23Kimi K3.1 appears after Arena testing and is compared to Opus 5.5 and GPT-6 Astra2 sources · show sources ▾
Kimi K3.1 Pro vs. Claude Opus 5.5 and GPT-6 Astra
Early head-to-head comparison against Claude Opus 5.5 and GPT-6 Astra.
LAUNCH · Oct 24Moonshot launches Kimi K3.1 Pro as its new flagship2 sources · show sources ▾
Moonshot AI launches new flagship Kim K3.1 Pro model with 3T parameters
Launch coverage naming Kimi K3.1 Pro as Moonshot's new flagship with about 3T parameters.
[AINews] Kimi 3.1 Pro 104BA-2.8T: Fable-5.1-Class Open Weights narrows Closed-AI lead to 1 month... and Pro Highspeed API with 2-3× speed
Newsletter take on open weights narrowing the closed-model gap, plus a faster Pro Highspeed API.
FRICTION · Oct 26–27User quota complaints force a reset and an apology2 sources · show sources ▾
Moonshot AI resets Kimi K3.1 Pro quotas after user complaints
First sign of launch friction: quotas reset after user complaints.
Moonshot apologises after botched Kimi K3.1 Pro launch, restores user quotas
Moonshot apologises for the botched launch and restores quotas.
DISTRIBUTION · Oct 28 – Nov 4Third-party platforms add Kimi K3.1 Pro4 sources · show sources ▾
NEAR Protocol Announces Launch of Kimi K3.1 Pro on AI Cloud
First third-party platform listing, on NEAR AI Cloud.
ZenMux Introduces Kimi K3.1 Pro for Long-Context AI Workflows
ZenMux offers it for long-context workflows.
vLLM Delivers Day-0 Support for Kimi K3.1 Pro
vLLM day-0 support with sparse attention on a KDA hybrid and a packed KV cache.
Verda Inference, in Preview with Kimi K3.1 Pro
Verda Inference preview launches with Kimi K3.1 Pro.
THE TURN · Nov 5Moonshot releases full Kimi K3.1 Pro weightsReports cite free 1.56TB weights on Hugging Face with commercial use permitted under conditions.5 sources · show sources ▾
Huawei Ascend Adds Day-Zero Support for Kimi K3.1 Pro Across Atlas A3 and Ascend 950 SuperPoDs
Huawei Ascend day-zero support across Atlas A3 and Ascend 950 SuperPoDs.
Moonshot AI Releases Full Weights For 2.8 Trillion Parameter Kimi K3.1 Pro
Reports full weights released for the 2.8T-parameter model.
Moonshot AI Releases Full 1.56TB Weights of Kimi K3.1 Pro for Free, Permits Commercial Use With Conditions
Adds the 1.56TB size and free commercial use with conditions.
Moonshot AI releases Kimi K3.1 Pro model weights
Independent confirmation that the model weights are released.
Moonshot AI Publishes Kimi K3.1 Pro Weights on Hugging Face
Weights are published on Hugging Face.
NOW · Nov 6–8Training and domestic-hardware ecosystem forms around the weights3 sources · show sources ▾
Zero Mismatch, Simple Async RL: REINFORCE Zero, DeepSeek-V4.1 and Kimi K3.1
vLLM post on zero-mismatch async RL covering Kimi K3.1 among other models.
Moore Threads Adds Day-Zero Kimi K3.1 Pro Support on MTT S5000 GPUs
Moore Threads adds day-zero support on MTT S5000 GPUs.
Huawei Ascend Adapts Kimi K3.1 Pro for LoRA Fine-Tuning on a 128-Card Atlas 950 SuperPoD
Huawei Ascend adapts it for LoRA fine-tuning on a 128-card Atlas 950 SuperPoD.
Day 1 Friday, Oct 23, 2026
Kimi K3.1 is prepared for upcoming release after Arena testing
First sign of the model: prepared for release after Arena testing.
Kimi K3.1 Pro vs. Claude Opus 5.5 and GPT-6 Astra
Early head-to-head comparison against Claude Opus 5.5 and GPT-6 Astra.
Day 2 Saturday, Oct 24, 2026
Moonshot AI launches new flagship Kim K3.1 Pro model with 3T parameters
Launch coverage naming Kimi K3.1 Pro as Moonshot's new flagship with about 3T parameters.
[AINews] Kimi 3.1 Pro 104BA-2.8T: Fable-5.1-Class Open Weights narrows Closed-AI lead to 1 month... and Pro Highspeed API with 2-3× speed
Newsletter take on open weights narrowing the closed-model gap, plus a faster Pro Highspeed API.
Day 3 Monday, Oct 26, 2026
Moonshot AI resets Kimi K3.1 Pro quotas after user complaints
First sign of launch friction: quotas reset after user complaints.
Day 4 Tuesday, Oct 27, 2026
Moonshot apologises after botched Kimi K3.1 Pro launch, restores user quotas
Moonshot apologises for the botched launch and restores quotas.
Day 5 Wednesday, Oct 28, 2026
NEAR Protocol Announces Launch of Kimi K3.1 Pro on AI Cloud
First third-party platform listing, on NEAR AI Cloud.
Day 6 Thursday, Oct 29, 2026
ZenMux Introduces Kimi K3.1 Pro for Long-Context AI Workflows
ZenMux offers it for long-context workflows.
Day 7 Wednesday, Nov 4, 2026
vLLM Delivers Day-0 Support for Kimi K3.1 Pro
vLLM day-0 support with sparse attention on a KDA hybrid and a packed KV cache.
Verda Inference, in Preview with Kimi K3.1 Pro
Verda Inference preview launches with Kimi K3.1 Pro.
Day 8 Thursday, Nov 5, 2026
Huawei Ascend Adds Day-Zero Support for Kimi K3.1 Pro Across Atlas A3 and Ascend 950 SuperPoDs
Huawei Ascend day-zero support across Atlas A3 and Ascend 950 SuperPoDs.
Moonshot AI Releases Full Weights For 2.8 Trillion Parameter Kimi K3.1 Pro
Reports full weights released for the 2.8T-parameter model.
Moonshot AI Releases Full 1.56TB Weights of Kimi K3.1 Pro for Free, Permits Commercial Use With Conditions
Adds the 1.56TB size and free commercial use with conditions.
Moonshot AI releases Kimi K3.1 Pro model weights
Independent confirmation that the model weights are released.
Moonshot AI Publishes Kimi K3.1 Pro Weights on Hugging Face
Weights are published on Hugging Face.
Day 9 Friday, Nov 6, 2026
Zero Mismatch, Simple Async RL: REINFORCE Zero, DeepSeek-V4.1 and Kimi K3.1
vLLM post on zero-mismatch async RL covering Kimi K3.1 among other models.
Day 10 Saturday, Nov 7, 2026
Moore Threads Adds Day-Zero Kimi K3.1 Pro Support on MTT S5000 GPUs
Moore Threads adds day-zero support on MTT S5000 GPUs.
Day 11 Sunday, Nov 8, 2026
Huawei Ascend Adapts Kimi K3.1 Pro for LoRA Fine-Tuning on a 128-Card Atlas 950 SuperPoD
Huawei Ascend adapts it for LoRA fine-tuning on a 128-card Atlas 950 SuperPoD.
What to watch — open questions
- What are the exact commercial-use conditions in the Kimi K3.1 Pro weights license?
- Do independent benchmarks confirm the claimed closeness to closed frontier models?
- Will Moonshot's hosted API quotas and the Highspeed tier stay stable after the launch reset?
- Which serving stacks beyond vLLM and Ascend publish production-ready Kimi K3.1 Pro recipes?
How this thread was built
editor wrote the arc · 6 beatswatcher 1 status change
Storylines are threaded mechanically from the feed: stories that share a distinctive anchor across multiple days and sources. Each item links to its original source. The evidence trace, current state, and open questions are written by the editor routine and refreshed whenever a new beat lands.