Model comparison
GPT-5.4 Pro vs Kimi K2.5
Head-to-head evidence from 6 shared benchmark results across 5 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Public leaderboard positions: GPT-5.4 Pro #44 (Estimated); Kimi K2.5 #54 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. GPT-5.4 Pro and Kimi K2.5 share 6 comparable benchmark results. 5 of 8 categories are comparable. 6 results are unique to GPT-5.4 Pro; 57 to Kimi K2.5.
Updated July 23, 2026- Shared results
- 6
- GPT-5.4 Pro only
- 6
- Kimi K2.5 only
- 57
- Comparable categories
- 5 / 8
Pick GPT-5.4 Pro if you want the stronger benchmark profile. Kimi K2.5 only becomes the better choice if mathematics is the priority or you want the cheaper token bill.
Confidence note. This is a partial-evidence comparison with 6 shared benchmark results across 5 evidence categories; 5 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
GPT-5.4 Pro has the cleaner BenchAlign overall profile here, landing at 60.89 versus 59.66. It is a real lead, but still close enough that category-level strengths matter more than the headline number.
GPT-5.4 Pro's sharpest advantage is in agentic, where it averages 89.3 against 55. The single biggest benchmark swing on the page is FrontierMath v2 (Tier 4), 37.500% to 4.200%. Kimi K2.5 does hit back in mathematics, so the answer changes if that is the part of the workload you care about most.
GPT-5.4 Pro is also the more expensive model on tokens at $30.00 input / $180.00 output per 1M tokens, versus $0.60 input / $3.00 output per 1M tokens for Kimi K2.5. That is roughly 60.0x on output cost alone. GPT-5.4 Pro is the reasoning model in the pair, while Kimi K2.5 is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use. GPT-5.4 Pro gives you the larger context window at 1.05M, compared with 256K for Kimi K2.5.
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | GPT-5.4 Pro | Δ | Kimi K2.5 |
|---|---|---|---|
| Agentic | GPT-5.4 Pro89.3 | Margin← 34.3 | Kimi K2.555.0 |
| Reasoning | GPT-5.4 Pro83.3 | Margin← 22.3 | Kimi K2.561.0 |
| Multimodal | GPT-5.4 Pro94.0 | Margin← 15.5 | Kimi K2.578.5 |
| Math | GPT-5.4 Pro46.9 | Margin→ 13.7 | Kimi K2.560.6 |
| Knowledge | GPT-5.4 Pro58.7 | Margin← 1.8 | Kimi K2.556.9 |
| Coding | GPT-5.4 ProNot measured | MarginNo overlap | Kimi K2.559.4 |
| Multilingual | GPT-5.4 ProNot measured | MarginNo overlap | Kimi K2.582.3 |
| Inst. Following | GPT-5.4 ProNot measured | MarginNo overlap | Kimi K2.593.9 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
FrontierMath v2 (Tier 4)
MathA 37.500%B 4.200%Winner: GPT-5.4 ProΔ 33.3FrontierMath v2 (Tier 4): GPT-5.4 Pro scored 37.500%; Kimi K2.5 scored 4.200%. GPT-5.4 Pro wins this benchmark. - Source ↗
BrowseComp
AgenticA 89.3%B 60.6%Winner: GPT-5.4 ProΔ 28.7BrowseComp: GPT-5.4 Pro scored 89.3%; Kimi K2.5 scored 60.6%. GPT-5.4 Pro wins this benchmark. - Source ↗
HLE
KnowledgeA 58.7%B 30.1%Winner: GPT-5.4 ProΔ 28.6HLE: GPT-5.4 Pro scored 58.7%; Kimi K2.5 scored 30.1%. GPT-5.4 Pro wins this benchmark. - Source ↗
FrontierMath v2 (Tiers 1-3)
MathA 50.000%B 27.900%Winner: GPT-5.4 ProΔ 22.1FrontierMath v2 (Tiers 1-3): GPT-5.4 Pro scored 50.000%; Kimi K2.5 scored 27.900%. GPT-5.4 Pro wins this benchmark. - Source ↗
MMMU-Pro
MultimodalA 94%B 78.5%Winner: GPT-5.4 ProΔ 15.5MMMU-Pro: GPT-5.4 Pro scored 94%; Kimi K2.5 scored 78.5%. GPT-5.4 Pro wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | GPT-5.4 Pro | Kimi K2.5 | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | GPT-5.4 Pro$30 input / $180 output | Kimi K2.5$0.6 input / $3 output | Kimi K2.5 has the lower combined listed price. |
| Generation speedtokens per second | GPT-5.4 Pro74 tok/s | Kimi K2.545 tok/s | GPT-5.4 Pro has the higher measured throughput. |
| First-answer latencyseconds to first token | GPT-5.4 Pro151.79 s | Kimi K2.52.38 s | Kimi K2.5 reaches the first token sooner. |
| Context windowmaximum listed tokens | GPT-5.4 Pro1.05M | Kimi K2.5256K | GPT-5.4 Pro lists the larger context window. |
Benchmark Deep Dive
AgenticGPT-5.4 Pro wins19 benchmarks
| Benchmark | GPT-5.4 Pro | Kimi K2.5 | Result |
|---|---|---|---|
| BrowseCompSource | 89.3% | 60.6% | GPT-5.4 Pro leads |
| Terminal-Bench 2.0Source | — | 50.8% | Not comparable |
| Claw-EvalSource | — | 52.3% | Not comparable |
| QwenClawBenchSource | — | 54.3% | Not comparable |
| τ³-bench resultsSource | — | 65.7% | Not comparable |
| DeepSearchQASource | — | 77.1% | Not comparable |
| DeepPlanningSource | — | 14.4% | Not comparable |
| ToolathlonSource | — | 27.8% | Not comparable |
| MCP AtlasSource | — | 29.5% | Not comparable |
| MCP-TasksSource | — | 59.1% | Not comparable |
| WideResearchSource | — | 72.7% | Not comparable |
| τ²-bench resultsSource | — | 95.9% | Not comparable |
| APEX-Agents-AASource | — | 11.5% | Not comparable |
| Gert LabsSource | — | 45.88% | Not comparable |
| ResearchClawBenchSource | — | 14.0% | Not comparable |
| JobBenchSource | — | 8.7% | Not comparable |
| AA Agentic IndexSource | — | 21.7% | Not comparable |
| GDPval-AASource | — | 25.4% | Not comparable |
| GDPval-AASource | — | 1009 | Not comparable |
Coding10 benchmarks
| Benchmark | GPT-5.4 Pro | Kimi K2.5 | Result |
|---|---|---|---|
| SWE-bench VerifiedSource | — | 76.8% | Not comparable |
| SWE-bench Verified*Source | — | 70.8% | Not comparable |
| LiveCodeBench v6Source | — | 85.0% | Not comparable |
| SWE-bench ProSource | — | 50.7% | Not comparable |
| SWE MultilingualSource | — | 73% | Not comparable |
| SWE-RebenchSource | — | 58.5% | Not comparable |
| React Native EvalsSource | — | 77.2% | Not comparable |
| SciCodeSource | — | 48.7% | Not comparable |
| AA-SciCodeSource | — | 49.0% | Not comparable |
| AA Coding IndexSource | — | 46.8% | Not comparable |
ReasoningGPT-5.4 Pro wins4 benchmarks
KnowledgeGPT-5.4 Pro wins15 benchmarks
| Benchmark | GPT-5.4 Pro | Kimi K2.5 | Result |
|---|---|---|---|
| HLESource | 58.7% | 30.1% | GPT-5.4 Pro leads |
| FrontierScienceSource | 36.7% | — | Not comparable |
| FrontierScience ResearchSource | 36.7% | — | Not comparable |
| HLE w/o toolsSource | 42.7% | — | Not comparable |
| GPQASource | — | 87.6% | Not comparable |
| GPQA-DSource | — | 87.6% | Not comparable |
| SuperGPQASource | — | 69.2% | Not comparable |
| MMLU-ProSource | — | 87.1% | Not comparable |
| MMLU-Pro (Arcee)Source | — | 87.1% | Not comparable |
| Artificial Analysis Intelligence IndexSource | — | 35.4% | Not comparable |
| AA-GPQA DiamondSource | — | 87.9% | Not comparable |
| AA-HLESource | — | 29.4% | Not comparable |
| AA-Omniscience IndexSource | — | -8.1% | Not comparable |
| AA-Omniscience AccuracySource | — | 34.3% | Not comparable |
| AA-Omniscience Hallucination RateSource | — | 64.6% | Not comparable |
MathKimi K2.5 wins11 benchmarks
| Benchmark | GPT-5.4 Pro | Kimi K2.5 | Result |
|---|---|---|---|
| IPhO 2025 (Theory)Source | 93.5% | — | Not comparable |
| FrontierMath (legacy)Source | 50% | — | Not comparable |
| FrontierMath v2 (Tiers 1-3)Source | 50.000% | 27.900% | GPT-5.4 Pro leads |
| FrontierMath v2 (Tier 4)Source | 37.500% | 4.200% | GPT-5.4 Pro leads |
| AIME 2025Source | — | 96.1% | Not comparable |
| AIME26Source | — | 95.8% | Not comparable |
| AIME25 (Arcee)Source | — | 96.3% | Not comparable |
| HMMT Feb 2025Source | — | 95.4% | Not comparable |
| HMMT Nov 2025Source | — | 91.1% | Not comparable |
| HMMT Feb 2026Source | — | 87.1% | Not comparable |
| MMAnswerBenchSource | — | 81.8% | Not comparable |
Multilingual2 benchmarks
MultimodalGPT-5.4 Pro wins6 benchmarks
Frequently Asked Questions (6)
Which is better, GPT-5.4 Pro or Kimi K2.5?
GPT-5.4 Pro is ahead on BenchLM's BenchAlign leaderboard, 60.89 to 59.66. The biggest single separator in this matchup is FrontierMath v2 (Tier 4), where the scores are 37.500% and 4.200%.
Which is better for knowledge tasks, GPT-5.4 Pro or Kimi K2.5?
GPT-5.4 Pro has the edge for knowledge tasks in this comparison, averaging 58.7 versus 56.9. Inside this category, HLE is the benchmark that creates the most daylight between them.
Which is better for math, GPT-5.4 Pro or Kimi K2.5?
Kimi K2.5 has the edge for math in this comparison, averaging 60.6 versus 46.9. Inside this category, FrontierMath v2 (Tier 4) is the benchmark that creates the most daylight between them.
Which is better for reasoning, GPT-5.4 Pro or Kimi K2.5?
GPT-5.4 Pro has the edge for reasoning in this comparison, averaging 83.3 versus 61. Inside this category, CritPt is the benchmark that creates the most daylight between them.
Which is better for agentic tasks, GPT-5.4 Pro or Kimi K2.5?
GPT-5.4 Pro has the edge for agentic tasks in this comparison, averaging 89.3 versus 55. Inside this category, BrowseComp is the benchmark that creates the most daylight between them.
Which is better for multimodal and grounded tasks, GPT-5.4 Pro or Kimi K2.5?
GPT-5.4 Pro has the edge for multimodal and grounded tasks in this comparison, averaging 94 versus 78.5. Inside this category, MMMU-Pro is the benchmark that creates the most daylight between them.
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
Related Comparisons
Explore More
Choose a model with this week’s evidence
Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.
One email each week. Unsubscribe anytime.