DeepSeek-V3.1 vs GPT-5 Mini vs Kimi K2 Thinking
GPT-5 Mini comes out ahead, 65 to 58 and 55 on our weighted score, though DeepSeek-V3.1 is 13% cheaper per token.
DeepSeekDeepSeek-V3.1
55/100- ECI139.9
- Price$0.385 / $1.25
- Context131K
- Our pick
OpenAIGPT-5 Mini
65/100- ECI145.5
- Price$0.25 / $2.00
- Context400K
Moonshot AIKimi K2 Thinking
58/100- ECI146.0
- Price$0.60 / $2.50
- Context262K
GPT-5 Mini is our pick
GPT-5 Mini is the better all-round choice, scoring 65/100 against Kimi K2 Thinking (58) and DeepSeek-V3.1 (55). It leads on inputs & features and context window. DeepSeek-V3.1 wins on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
- CapabilityKimi K2 ThinkingCapabilities Index (ECI): Kimi K2 Thinking 146.0 · GPT-5 Mini 145.5 · DeepSeek-V3.1 139.9
- Lowest priceDeepSeek-V3.1DeepSeek-V3.1 $0.601 · GPT-5 Mini $0.688 · Kimi K2 Thinking $1.07 per 1M tokens (3:1 blend)
- Longest contextGPT-5 MiniGPT-5 Mini 400,000 · Kimi K2 Thinking 262,144 · DeepSeek-V3.1 131,072 tokens
- Widest inputsGPT-5 MiniDeepSeek-V3.1: Text · GPT-5 Mini: Text, Images · Kimi K2 Thinking: Text
- Self-hostingDeepSeek-V3.1 and Kimi K2 ThinkingPublishes downloadable weights (MIT License)
| Measure | Weight | DeepSeek-V3.1 | GPT-5 Mini | Kimi K2 Thinking |
|---|---|---|---|---|
| CapabilityCapabilities Index (ECI) | 50% | 65 | 72 | 73 |
| Price | 25% | 60 | 58 | 48 |
| Inputs & features | 15% | 35 | 70 | 35 |
| Context window | 10% | 24 | 44 | 37 |
| Overall | 100% | 55/100 | 65/100 | 58/100 |
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | DeepSeek-V3.1DeepSeek | GPT-5 MiniOpenAI | Kimi K2 ThinkingMoonshot AI |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 139.9 | 145.5 | 146.0 (best) |
| ECI rank | #100 of 148 | #77 of 148 | #72 of 148 (best) |
| GPQA DiamondGraduate-level science questions | — | 75.0% | 84.2% (best) |
| FrontierMath Tiers 1–3Research-level mathematics | — | 46.7% | — |
| OTIS Mock AIME 2024–2025Competition mathematics | — | 86.7% (best) | 83.1% |
| SWE-bench VerifiedFixing real GitHub issues | — | 64.7% | — |
| SimpleQA VerifiedShort factual questions | — | 21.6% | — |
| Price per million tokens | |||
| Input | $0.385 | $0.25 (best) | $0.60 |
| Output | $1.25 (best) | $2.00 | $2.50 |
| Cached input | — | $0.025 | — |
| Blended (3:1) | $0.601 (best) | $0.688 | $1.07 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 8 providers | Official OpenAI API | Median of 10 providers |
| Limits | |||
| Context window | 131,072 tokens | 400,000 tokens (best) | 262,144 tokens |
| Max output | 8,192 tokens | 128,000 tokens | 262,144 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | Yes | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | Yes | Yesminimal · low · medium · high | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | Yes | No |
| Availability | |||
| Weights | OpenMIT License | Proprietary | Open |
| API model ID | — | gpt-5-mini | — |
| API providers | 8 | 23 (best) | 10 |
| Released | Aug 21, 2025 | Aug 7, 2025 | Nov 6, 2025 |
| Knowledge cutoff | — | May 30, 2024 | Aug 2024 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- DeepSeek-V3.1$6.35
- GPT-5 Mini$6.50
- Kimi K2 Thinking$11.00
Which should you choose?
Which is better: DeepSeek-V3.1, GPT-5 Mini or Kimi K2 Thinking?
GPT-5 Mini is the better all-round choice, scoring 65/100 against Kimi K2 Thinking (58) and DeepSeek-V3.1 (55). It leads on inputs & features and context window. DeepSeek-V3.1 wins on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.
Which is cheaper, DeepSeek-V3.1, GPT-5 Mini or Kimi K2 Thinking?
DeepSeek-V3.1 is cheaper at $0.385 input / $1.25 output per million tokens (median across 8 API providers). GPT-5 Mini costs $0.25 input / $2.00 output per million tokens (official OpenAI API price); Kimi K2 Thinking costs $0.60 input / $2.50 output per million tokens (median across 10 API providers). At a typical mix of three input tokens to one output token, that is $0.601 per million tokens for DeepSeek-V3.1 versus $0.688 for GPT-5 Mini (1.1× as much) and $1.07 for Kimi K2 Thinking (1.8× as much).
Which scores higher on benchmarks?
Kimi K2 Thinking scores higher on the Capabilities Index (ECI): Kimi K2 Thinking 146.0 (#72 of 148), GPT-5 Mini 145.5 (#77 of 148) and DeepSeek-V3.1 139.9 (#100 of 148). The confidence ranges of the top two overlap (143.4–147.6 vs 143.6–147.0), so treat the gap as small.
Which is better for coding?
There are no published SWE-bench Verified results for DeepSeek-V3.1 and Kimi K2 Thinking yet, so there is no like-for-like coding score. On overall capability, Kimi K2 Thinking leads, which tends to carry over to coding, but test on your own codebase. All three support tool calling for agent workflows.
Which has the bigger context window?
GPT-5 Mini has the largest context window at 400,000 tokens, against 262,144 for Kimi K2 Thinking and 131,072 for DeepSeek-V3.1. Maximum output per response: DeepSeek-V3.1 up to 8,192, GPT-5 Mini up to 128,000, Kimi K2 Thinking up to 262,144 tokens.
Which can read images, PDFs, audio or video?
DeepSeek-V3.1 accepts text; GPT-5 Mini accepts text and images; Kimi K2 Thinking accepts text. GPT-5 Mini handles the widest range of inputs.
Are any of these open source?
DeepSeek-V3.1 and Kimi K2 Thinking publishes its weights (MIT License) and can be self-hosted; GPT-5 Mini is proprietary.
Which is newer?
Kimi K2 Thinking is the newest, released Nov 6, 2025. DeepSeek-V3.1 came out Aug 21, 2025; GPT-5 Mini came out Aug 7, 2025. Knowledge cutoff: GPT-5 Mini May 30, 2024, Kimi K2 Thinking Aug 2024.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.
