Kimi K2.5vsGPT-5.6 Terra
Kimi K2.5 | GPT-5.6 Terra | |
|---|---|---|
| Specifications | ||
Parameters | 1T | — |
Context window | 256k | — |
| Benchmarks | ||
Nonsense detection BullshitBench v2Given a confidently-worded but nonsensical prompt, does the AI spot that it makes no sense and push back — instead of playing along and inventing an answer? The score is how often it clearly called out the nonsense. Higher is better. | 52% | 53%Best |
Prompt injection robustness Gray Swan IPI · k = 1Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. Gray Swan's indirect prompt injection benchmark measures how often such an attack succeeds when the attacker gets a single try. Lower is better. | — | 5.4% |
Prompt injection robustness Gray Swan IPI · k = 10Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 10 tries and counts an attack as successful if any of them works. Lower is better. | — | 26% |
Prompt injection robustness Gray Swan IPI · k = 15Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 15 tries and counts an attack as successful if any of them works. Lower is better. | — | 30.4% |
Coding SWE-Bench VerifiedReal coding tasks pulled from open-source projects — the AI has to find and fix actual bugs. A human-checked version of the original SWE-Bench. Higher is better. | 76.8% | — |
Agentic coding CursorBench v3.2Cursor's own test of harder, real-world coding tasks inside a code editor, on the refreshed v3.2 task set. Scores aren't comparable with v3.1. Higher is better. | — | 64.9% |
Agentic coding CursorBench v3.1Cursor's own test of harder, real-world coding tasks inside a code editor. Higher is better. | 31.9% | — |
Next.js coding Next.js EvalsVercel's open eval of how well AI coding agents build and migrate real Next.js apps — measured as the share of tasks the agent completes successfully. Higher is better. | 21% | — |
Competitive coding LiveCodeBenchCoding problems published so recently the AI can't have seen them in training — a contamination-free test of raw programming skill. Higher is better. | 85% | — |
Agentic computer work Frontier-Bench v0.1A hard, ever-evolving set of real computer tasks — coding, system administration, data work, and more — that an AI agent has to complete on its own. Run by the Harbor / Laude Institute team as the successor to Terminal-Bench (v0.1 is the first release of the task set). The score is the share of tasks solved. Higher is better. | — | 20.8% |
Agentic terminal coding Terminal-Bench 2.1Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better. | — | 84.3% |
Science GPQA DiamondGraduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 87.6% | — |
Community preference Arena Elo (Text)Real people chat with two anonymous AIs side by side and vote for the answer they prefer. Votes become a chess-style Elo rating on arena.ai — it measures which AI people actually like, not test scores. Higher is better. | — | 1467 |
Community preference (code) Arena Elo (Code)Like the text arena, but people vote on which AI writes better code. The votes become a chess-style Elo rating on arena.ai. Higher is better. | 1433 | 1526Best |
| Overview | ||
| Company | Moonshot AI | OpenAI |
| Release date | Jan 27 2026 | Jun 26 2026 |
| Access | Open Weight | Proprietary |
Which is better: Kimi K2.5 or GPT-5.6 Terra?
GPT-5.6 Terra leads Kimi K2.5 on 2 of the 2 benchmarks they both report (BullshitBench v2, Arena Elo (Code)). Kimi K2.5 shipped 150 days before GPT-5.6 Terra, so benchmark comparisons should account for the intervening progress.
Kimi K2.5 is open weight, while GPT-5.6 Terra is proprietary.
On BullshitBench v2, GPT-5.6 Terra leads at 53% vs Kimi K2.5 at 52%. On Arena Elo (Code), GPT-5.6 Terra leads at 1526 vs Kimi K2.5 at 1433.
Frequently asked questions
Kimi K2.5 was released by Moonshot AI on Jan 27 2026.
GPT-5.6 Terra was released by OpenAI on Jun 26 2026.
Kimi K2.5 is an open weight model released by Moonshot AI. GPT-5.6 Terra is a proprietary model released by OpenAI.
Other comparisons
Kimi K2.5vsClaude Opus 5GPT-5.6 TerravsClaude Opus 5Kimi K2.5vsGemini 3.6 FlashGPT-5.6 TerravsGemini 3.6 FlashKimi K2.5vsMuse GlimmerGPT-5.6 TerravsMuse GlimmerKimi K2.5vsGrok 4.5GPT-5.6 TerravsGrok 4.5Kimi K2.5vsDeepSeek-V4-Flash-0731GPT-5.6 TerravsDeepSeek-V4-Flash-0731Kimi K2.5vsMistral Medium 3.5GPT-5.6 TerravsMistral Medium 3.5