Today for AI
HOT RADAR
llmHEAT 0.0°

AI CURATED HOT STORY · 10/24/2026

Moonshot AI Releases Kimi K3.1 Pro, Narrowing Gap with Top Closed Models

21 reports archived0 independent sourcesupdated 10/24/2026, 22:57:12
Synthesis & Latest Updates

Moonshot AI has released Kimi K3.1 Pro, its most powerful open-weight model, marking a significant breakthrough in narrowing the performance gap between open-source and top-tier closed models. Built on a 2.8T total parameter (104B active) MoE architecture with a 1M token context window, the model scores 55 on the Artificial Analysis Intelligence Index, ranking fourth globally and improving by 11 points over its predecessor. Upon weight release on November 4, it will become the highest-scoring open-weight model, reducing the gap with closed flagships from 12 to just 3 points.

Cross-verified data highlights exceptional capabilities in coding, 3D generation, and financial research. The model’s Terminal-Bench score surged from 13% to 67%, surpassing Claude Opus 5.5 in specific metrics, while ranking 8th on Code Arena WebDev with performance comparable to Claude Opus-5 at 70% lower cost. Hallucination rates dropped by 65%, and it secured first place in OfficeQA Pro v2 for financial multi-step calculations. A new Highspeed version offers 2-3x faster inference for real-time interactions, though the standard version suffers from slower output speeds and verbosity.

This release reshapes the developer ecosystem by delivering near-closed-model intelligence at aggressive pricing ($2/$10 per million tokens), significantly lowering barriers for high-quality code generation and specialized modeling. By compressing the performance gap, Kimi K3.1 Pro sets a new open-source benchmark, compelling closed vendors to rethink pricing strategies and accelerating the democratization of advanced AI capabilities.

LATEST/Kimi K3.1 Pro is live via API, with full weights scheduled for release on November 4. Current discussions focus on the standard version's slower inference speed and verbose outputs, mitigated by the Highspeed variant offering 2-3x acceleration. The community awaits confirmation of its status as the strongest open model, particularly regarding its potential to displace competitors like DeepSeek in coding and agentic tasks.

Hourly heat curve

133 fully observed hours
Now
0.0
Peak
219.808:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 21 reports · Latest first
  2. IT之家 · 智能时代T2·78 pts

    Tencent WorkBuddy Integrates Kimi K3.1 Pro with 1M Context and High-Speed Variant

    • WorkBuddy integrates Kimi K3.1 Pro with 1M context support and launches a high-speed variant with 2-3x faster output.
    • The 3-trillion-parameter Kimi K3.1 Pro performs exceptionally well on Terminal-Bench and BenchCAD, approaching or surpassing GPT-6 and Claude Opus 5.5.
    • The model achieves State-of-the-Art (SoTA) results on OfficeQA Pro v2 and SpreadsheetBench v2, demonstrating strong end-to-end financial workflow delivery capabilities.
  6. IT之家 · 智能时代T2·88 pts

    Moonshot AI Releases Kimi K3.1 Pro, Outperforming GPT-6 in Coding Benchmarks

    • Kimi K3.1 Pro scores 65.7% on Terminal-Bench 4.0, surpassing GPT-6 Astra (58.2%) and approaching Claude Opus 5.5.
    • Enhanced 3D creation and spatial reasoning allow the model to directly write and execute scripts for professional tools like Blender and CAD without relying on MCP connectors.
    • Available in standard and high-speed versions (2-3x faster output) with clear API pricing; full model weights will be released by November 4.
  7. Artificial Analysis (@ArtificialAnlys)T1·78 pts

    Kimi K3.1 Pro Tops Open-Source Cyber Defense Benchmark

    • Kimi K3.1 Pro ranks #1 among openly-available models with a 64% score, trailing only the restricted-access GPT-6 Sol.
    • Cyber defense capabilities were evaluated across three benchmarks: CWE-Bench-AA, DeepsecBench-AA, and CyberGym-E2E-AA.
    • While per-token pricing is 33% lower than Kimi K3, total task costs rose by 15% due to increased input token consumption.
  8. Simon Willison 博客T2·88 pts

    Moonshot Releases Kimi K3.1 Pro with Open Weights Coming Soon

    • Kimi K3.1 Pro is live on API with input at $2/million tokens and output at $10/million; a 'Highspeed' version offers 2-3x speed at double the price.
    • Scores 55 on Artificial Analysis, trailing only Claude Opus 5.5 (58) and surpassing GPT-6 Astra and Gemini 4 Argon.
    • Moonshot promises full model weights release by Nov 4, 2026, making it the first near-frontier model to go fully open source.
  9. Kimi 博客 (月之暗面)T1·88 pts

    Moonshot AI Releases Kimi K3.1 Pro: A 3T-Parameter Open Model with 1M Context

    • Kimi K3.1 Pro is the first 3T-class open-weight model with a 1-million-token context window.
    • Significant performance gains in coding, 3D generation, and financial research agentic tasks.
    • The Highspeed variant offers 2-3x faster inference compared to the standard version.
    • Full model weights will be officially released on November 4.
  10. Rohan Paul (@rohanpaul_ai)T2·85 pts

    Moonshot Deliberately Weakens Kimi K3.1 Pro Hacking Skills Before Release

    • Moonshot removed cyber-exploit data during mid-training and RL, reducing internal xdev-bench-hard pass@3 from 90.5% to 81.0%
    • The filtered model still solves 81% of exploitation tasks vs. 33.3% for K3, indicating general coding gains carry over into cyber work
    • Hallucination rates in tool-use scenarios dropped significantly from 23.4% (K3) to 8.3%, though remaining higher than GPT-6 Astra's 5.5%
  11. Rohan Paul (@rohanpaul_ai)T2·88 pts

    Moonshot AI Releases Kimi K3.1 Pro with Major Coding and Hallucination Improvements

    • Kimi K3.1 Pro scores 65.7% on Terminal-Bench 4.0, trailing Claude Opus 5.5 by only 0.7 points at a lower cost.
    • Removing cyber-exploit data from training reduced vulnerability exploitation success rates but maintained strong general coding capabilities.
    • Hallucination rates in tool-use scenarios dropped from 23.4% to 8.3%, with significant gains in simulated business operation tests.
  12. The DecoderT2·88 pts

    Moonshot AI Releases Kimi K3.1 Pro Closing Gap with Claude Opus 5.5

    • Kimi K3.1 Pro scores 55 on the AA Intelligence Index, ranking fourth globally behind Claude Opus 5.5, Claude Sonnet 5.5, and Muse Wonder Full.
    • The new model achieves an 11-point jump over its predecessor Kimi K3 by leveraging significantly more post-training compute.
    • Scheduled to release weights by November 4, 2026, it will shrink the gap between top open and proprietary models from 12 to just 3 points.
  13. OpenRouter (@OpenRouter)T1·88 pts

    Kimi K3.1 Pro Launches on OpenRouter with 1M Context and Multimodal Support

    • Kimi K3.1 Pro uses an MoE architecture with 2.8T total and 104B active parameters, supporting a 1M-token context.
    • Enhanced capabilities in multimodal tasks, long-horizon coding (Unreal/Unity/Godot), and 3D modeling (Blender/CAD).
    • Improved accuracy in financial modeling with a 65% reduction in hallucinations, matching top proprietary models.
    • Priced at $2 per million input tokens and $10 per million output tokens, with a high-speed variant available.
  14. Hacker News AIT2·68 pts

    Kimi K3.1 Pro Launches with Intelligence Score of 55 and High Cost-Efficiency

    • Kimi K3.1 Pro scores 55 on the Artificial Analysis Intelligence Index, placing it on the Pareto frontier with high intelligence relative to cost.
    • Aggressive pricing: $2.00 per million input tokens and $10.00 per million output tokens, with a 90% cache discount and ~$2.11 cost per task.
    • Performance trade-offs: Output speed is 73.5 tokens/s (below average), and the model exhibits very high verbosity.
  15. 宝玉 (@dotey)T2·85 pts

    Moonshot AI Releases Kimi K3.1 Pro with Focus on Finance and Coding

    • Kimi K3.1 Pro is a 2.8T-parameter MoE model with ~104B active parameters and 1M-token context support.
    • It matches or surpasses Claude Opus 5.5 and GPT-6 Astra in coding benchmarks like Terminal-Bench and SWE-Marathon.
    • Official claims indicate a 65% reduction in hallucinations compared to the previous generation, ranking first in financial multi-step reasoning (OfficeQA Pro v2).
    • Full model weights are scheduled for open-source release by November 4; web, app, and API access are currently live.
  16. Artificial Analysis (@ArtificialAnlys)T1·88 pts

    Kimi K3.1 Pro Launches: Ranks 4th on Intelligence Index, Open Weights Coming Soon

    • Significant boost in agentic benchmarks, with Terminal-Bench scores jumping from 13% to 67%
    • Pricing reduced by 33% compared to previous version ($2/$10/$0.20 per 1M input/output/cache hit tokens)
    • Will become the highest-scoring open-weights model upon release, narrowing the gap with proprietary leaders to 3 points
  21. Artificial AnalysisT1·85 pts

    Kimi K3.1 Pro Launches at #4 on Intelligence Index, Set to Lead Open Weights

    • Kimi K3.1 Pro scores 55 on the Intelligence Index, trailing only Claude Opus 5.5, Claude Sonnet 5.5, and Muse Wonder Full.
    • The model ranks third on the GDPval-AA agentic benchmark, gaining 312 Elo points over its predecessor.
    • Once weights are released, it will be the top open-weights model, reducing the gap between open and closed-source leaders from 12 to 3 points.