Updated October 10, 2026. Rank says GPT-5.2 is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.
Both of these models will change. Get the price, version and retirement notices for the pair, each with its source. Follow model changes
GPT-5.2 has the higher public point estimate, 60.33 versus 58.74. Their conditional score ranges overlap. These ranges do not establish rank confidence. 13 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Code generation, repair, and software-engineering tasks
Kimi K2.6
Kimi K2.6 has the higher public coding point estimate, 43.9 to 37.7, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Prompts that approach the documented context limit
GPT-5.2
GPT-5.2 has the larger documented context window.
1K fresh input + 500 output tokens
Kimi K2.6
Kimi K2.6 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
200K cached + 20K fresh input + 10K output tokens
Kimi K2.6
Kimi K2.6 has the lower estimated token cost for this stated workload. GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate.
50K fresh input + 3K output tokens
Kimi K2.6
Kimi K2.6 has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Tool use, computer use, and multi-step task completion
No clear pick
The like-for-like agentic result is a practical tie on the public lane (within 0.5 points).
Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.
The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.
Like-for-like · BenchAlign v5.8
Kimi K2.6 has the higher coding point estimate. Conditional score ranges do not establish rank confidence.
Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.
OSWorld-VerifiedAgentic
Normalized gap 25.8BrowseCompAgentic
Normalized gap 17.4FrontierMath v2 (Tier 4)Math
Normalized gap 4.2SWE-bench ProCoding
Normalized gap 3.0GPQAKnowledge
Normalized gap 1.9Each row shows the public-lane category score for both models: the BenchAlign v5.8 lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.
| Category | GPT-5.2 | Kimi K2.6 | Basis | Reading |
|---|---|---|---|---|
| Agentic | 43.6Supported · #58/123 | 43.7Supported · #57/123 | Like-for-likeBenchAlign v5.8 lane · 4 vs 12 public rows | Practical tie |
| Coding | 37.7Supported · #68/146 | 43.9Supported · #54/146 | Like-for-likeBenchAlign v5.8 lane · 3 vs 10 public rows | Kimi K2.6 leads · intervals overlap |
| Multimodal | 75.9#22/54 | 73.8#27/54 | Like-for-likeProvisional lane · 2 vs 2 weighted rows | GPT-5.2 leads |
| Knowledge | 59.6Supported · #46/177 | 59.8Supported · #45/177 | Like-for-likeBenchAlign v5.8 lane · 1 vs 5 public rows | Practical tie |
| Reasoning | 64.4Unranked · 4 rankable rows | Not ranked | Not comparableProvisional lane · 1 vs 0 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Instruction following | 92.6#19/127 | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | 57.9Unranked · 3 rankable rows | 72.2#1/7 | Not comparableProvisional lane · 2 vs 4 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.8) differ from the provisional-lane categories. Unranked scores sit on the provisional lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
1K fresh input + 500 output tokens
Kimi K2.6 has the lower modeled cost
Costs use the listed standard API rates.
50K fresh input + 3K output tokens
Kimi K2.6 has the lower modeled cost
Costs use the listed standard API rates.
200K cached + 20K fresh input + 10K output tokens
Kimi K2.6 has the lower modeled cost
GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate.
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
Maximum documented context; output-token limits may be lower.
GPT-5.2
400K
Kimi K2.6
256K
GPT-5.2
Not sourced
Kimi K2.6
Not sourced
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
GPT-5.2
Not published
Kimi K2.6
$0.16 per 1M cached input tokens
GPT-5.2
Not sourced
Kimi K2.6
Not sourced
GPT-5.2
Not sourced
Kimi K2.6
Not sourced
GPT-5.2
Not sourced
Kimi K2.6
Not sourced
GPT-5.2
Reasoning
Kimi K2.6
Reasoning
GPT-5.2
Proprietary
Kimi K2.6
Open Weight
GPT-5.2
Proprietary
Kimi K2.6
Open Weight
GPT-5.2
2025-12-11
Kimi K2.6
2026-04-20
GPT-5.2 has the higher public point estimate, 60.33 versus 58.74. Their conditional score ranges overlap. These ranges do not establish rank confidence. The page therefore keeps the decision tied to the specific documented workload.
Kimi K2.6 has the higher public coding point estimate, 43.9 to 37.7, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
The like-for-like agentic tasks row is a practical tie on the public lane, 43.6 against 43.7, inside the 0.5-point band BenchLM treats as level.
For the stated presets, chat costs $0.00875 on GPT-5.2 and $0.00295 on Kimi K2.6; repository review costs $0.1295 and $0.0595; the cache-heavy agent loop costs $0.525 and $0.091. GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate.
GPT-5.2 has the larger documented context window: 400K, compared with 256K.
Estimates at 50,000 req/day · 1000 tokens/req average.
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
BrowseComp
Kimi K2.6 leads this result
OSWorld-Verified
Kimi K2.6 leads this result
Gert Labs
Shared sourceKimi K2.6 leads this result
JobBench
Not directly comparable
Terminal-Bench 2.0
Not directly comparable
Toolathlon
Not directly comparable
MCP Atlas
Not directly comparable
Claw-Eval
Not directly comparable
DeepSearchQA
Not directly comparable
WideResearch
Not directly comparable
ResearchClawBench
Not directly comparable
OSWorld 2.0
Not directly comparable
Terminal-Bench 2.1 (Vals)
Not directly comparable
SWE-bench Verified
Kimi K2.6 leads this result
SWE-bench Pro
Kimi K2.6 leads this result
Vibe Code Bench
Shared sourceGPT-5.2 leads this result
LiveCodeBench v6
Not directly comparable
SWE Multilingual
Not directly comparable
SciCode
Not directly comparable
Terminal-Bench 2.0
Not directly comparable
cursorBench31
Not directly comparable
LiveCodeBench (Vals)
Not directly comparable
SWE-bench (Vals)
Not directly comparable
MMMU
Not directly comparable
MMMU-Pro
GPT-5.2 leads this result
RealWorldQA
Not directly comparable
OmniDocBench 1.5
Not directly comparable
Video-MME (with subtitle)
Not directly comparable
Video-MME (w/o subtitle)
Not directly comparable
MathVision
Kimi K2.6 leads this result
We-Math
Not directly comparable
DynaMath
Not directly comparable
SimpleVQA
Not directly comparable
CC-OCR
Not directly comparable
AI2D_TEST
Not directly comparable
CountBench
Not directly comparable
ERQA
Not directly comparable
VideoMMMU
Not directly comparable
MLVU (M-Avg)
Not directly comparable
CharXiv
GPT-5.2 leads this result
V*
Kimi K2.6 leads this result
MMMU-Pro w/ Python
Not directly comparable
GPQA
GPT-5.2 leads this result
GPQA-D
Not directly comparable
HLE
Not directly comparable
GPQA Diamond (Vals)
Not directly comparable
MMLU-Pro (Vals)
Not directly comparable
AIME 2025
Not directly comparable
FrontierMath v2 (Tiers 1-3)
Shared sourceGPT-5.2 leads this result
FrontierMath v2 (Tier 4)
Shared sourceGPT-5.2 leads this result
AIME26
Not directly comparable
HMMT Feb 2026
Not directly comparable
MMAnswerBench
Not directly comparable
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 5,500+ readers.
Last updated October 10, 2026
