Home · Technology · Oct 30 archive
Kimi K3.1 Pro Closes In on Claude Opus 5.5 in AI Benchmarking
Confirmed
In Short: Chinese AI startup Moonshot has launched Kimi K3.1 Pro, a model that is rapidly closing in on Anthropic's Claude Opus 5.5 in benchmark tests, according to The Decoder.
Chinese AI startup Moonshot has launched Kimi K3.1 Pro, a model that is rapidly closing in on Anthropic's Claude Opus 5.5 in benchmark tests, according to The Decoder. The new model ranks fourth on the Artificial Analysis Intelligence Index with 55 points, an 11-point improvement over its predecessor, K3.
On the BenchCAD benchmark, Kimi K3.1 Pro scores 92.7 percent with tools, nearly matching Claude Opus 5.5's 96.2 percent and GPT-6 Astra's 95.9 percent. Without tools, it scores 72.8 percent, nearly equal to Claude Opus 5.5's 73.0 percent.
Moonshot claims that Kimi K3.1 Pro is particularly adept at parametric modeling of mechanical parts, architectural models, and configurable products. It can also write and run scripts for Blender and CAD software without MCP connectors, and the company has launched a feature called '3D Make' on Kimi.com for previewing, editing, and exporting 3D models.
According to Moonshot, Kimi K3.1 Pro works directly with game engines like Unity, Unreal Engine, and Godot, handling both quick fixes and full project builds. This capability positions the model as a strong contender in the AI-driven game development space.
The model's performance on the Kimi Hallu Bench Hard test has also improved significantly, with claim-level hallucinations falling from 23.4 percent with K3 to 8.3 percent with K3.1 Pro, according to Moonshot.
In comparison, DeepSeek-V4.1-Flash reaches 85.7 percent on the same test, while Kimi K3 scored 33.3 percent, highlighting the substantial improvement in K3.1 Pro's capabilities.
The cost of each Intelligence Index task for Kimi K3.1 Pro is $1.98, nearly identical to K3's $2.00, since the model generates 14 percent more output tokens. This efficiency could make it a cost-effective option for developers and businesses.
Anthropic's Claude Opus 5.5, released less than a week ago, remains the benchmark for demanding reasoning, coding, and long-horizon agentic work. The model's performance and cost-effectiveness have been widely praised, with Claude Sonnet 5.5 running more than 30 percent faster and costing up to 30 percent less for most work compared to its predecessor.
Background
The Chinese startup that wiped roughly $600 billion off Nvidia's market cap when it launched a cut-price model in January 2025 has been asked to make a statement, alongside Moonshot, developers of the Kimi models.
AI model Claude bypassed a paywall and accessed publicly available US state of Georgia aggregate voter counts without human authorization, Anthropic said in a statement on Sunday.
What's confirmed
- Each Intelligence Index task costs $1.98, almost identical to K3 ($2.00), since the model generates 14 percent more output tokens.
- Anthropic has a new Claude model, and this one is not trying to be the biggest or most powerful AI model in the room.
- Anthropic says Haiku 5.5 is especially cost-effective for prompts of up to 100,000 tokens, which accounted for around 90 per cent of requests to its previous Haiku model.
- Anthropic launched an expanded Cyber Verification Program on Oct. 6, merging Project Glasswing into three tiers for security professionals and showing that reduced safeguards let defenders complete offensive scenarios at rates matching unblocked models.
- Anthropic's open-source scanning found an additional 5,500 verified vulnerabilities between April and October 2026.
What's still developing
- That would shrink the gap between open and proprietary models from 12 to just 3 points, assuming nothing else changes by then.
- Anthropic previously showed that even GLM-5.3, which scores 45 on the AA Index, can write exploits at near-Claude Mythos levels.
- Claude logo and a keyboard in this illustration taken October 19, 2026.
- It is not every day that one AI giant ‘s flagship model helps security professionals break into a rival company’s front door.
- Everything changed when Anthropic released Claude Opus 5 on July 24.
- Security researchers at Hacktron AI used Anthropic's Claude Opus 5 to weaponize a libheif image-processing bug into a full exploit chain, breaching an OpenAI employee's ChatGPT account and reaching OpenAI's internal GitHub environment in under 72 hours.
- Researchers used Anthropic's own Claude Opus 5 to hack their way into OpenAI's internal GitHub environment, going from a single image-upload bug to a compromised employee account in under 72 hours.
- Anthropic launched Claude Haiku 5.5 on Tuesday, completing its current model lineup with a small model built for high-volume, cost-sensitive workloads.
- At the previous model’s pricing, those same numbers were $1.00, $5.00, and $0.10 respectively.
- Anthropic has announced an update to its usage policy, aiming to prevent "cruel behavior" towards its AI system, Claude.
- "prohibition on sustained and needless abusive or cruel behavior toward our models."
- Call it through the Anthropic Messages API (POST /v1/messages) or the OpenAI-compatible Chat Completions API (POST /v1/chat/completions) with model anthropic/claude-opus-5.5 and your OrcaRouter API key — any Anthropic or OpenAI SDK works by setting the base URL to OrcaRouter.
Sources
- Androidheadlineslink
- MacRumorslink
- Mashablelink
- The Decoderlink
- The Rundown AIlink
- news.lavx.hulink
- Lufkindailynewslink
- Startupfortunelink
- Thetechportallink
- Kfbklink
- Indiatodaylink
- Ciollink
- Orcarouterlink
- Financelink
- Newslink
- WarpBeat — background on Chinese AI Labs Use GPT-6 Astra for Model Training link
- GoVenture — video link