MoonshotAI
·Released on Jan 27, 2026Kimi K2.5 API Benchmarks, Pricing & Provider Data
Compare Kimi K2.5 with another model
Choose a model to open its comparison page.
LLM
Kimi K2.5 benchmark, API pricing, and provider data cover undefined API provider} other undefined API providers}}, with prices starting at $0.0040/M. Kimi K2.5 free API options are available from undefined provider} other undefined providers}}. The page also shows measured API speed and first-token latency.
Moonshot Kimi K2.5 is an open-weight multimodal agent model with native vision and text input, strong coding performance, and a 256K context window.
Quality
#87of 112
49.0
LMSpeed score
Speed
144char/s
11.93 s
Cost
#96of 186
$0.0040/ 1M · 8:1 in:out
$0.0006 in · $0.0034 out
Category Performance
Observed-capability estimates with uncertainty reported separately from independent benchmark families.
- Coverage
- 8 / 8
- Methodology
- V3.0
#1Reasoning55.5RatedGlobal rank #173/4 Measured dimensions
80% interval: 47.2–63.8
abstract logicPrior only
scientific causal56.3
critpt · critpt / gpqa · gpqa / hle · hle
multistep constraints58.5
mmlu_pro · mmlu_pro
evidence verification51.8
lcr · lcr / longbench · long_bench_v2
#2Instruction following54.1Estimated2/4 Measured dimensions
80% interval: 42.2–66.0
constraint following55.6
ifeval · ifeval
structured outputPrior only
novel instruction generalization52.7
ifbench · aa_if_bench
long multiturn instructionPrior only
#3Multimodal52.4Estimated2/4 Measured dimensions
80% interval: 40.4–64.4
perception ocrPrior only
document spatialPrior only
visual reasoning50.5
mmmu_pro · mmmu_pro
video action54.3
video_mmmu · video_mmmu
#4Math49.1Estimated3/4 Measured dimensions
80% interval: 40.0–58.1
foundational mathPrior only
competition math51.7
aime · aime2025 / aime · aime2026 / hmmt · hmmt_feb2025 / hmmt · hmmt_feb2026 / hmmt · hmmt_nov2025
proof frontier50.7
frontiermath · frontier_math_v2_tier4 / frontiermath · frontier_math_v2_tiers13
applied tool math44.8
mmanswer · mm_answer_bench
#5Coding47.3RatedGlobal rank #303/4 Measured dimensions
80% interval: 39.6–55.1
code generation50.7
livecodebench · live_code_bench_v6 / scicode · sci_code
repository engineering43.7
react_native_bench · react_native_evals / swe_pro · swe_pro / vibecode · vibe_code_bench
debugging testing47.6
swe_multilingual · benchlm_coding_swe_multilingual / swe_rebench · swe_rebench / swe_verified · swe_verified
tooling qualityPrior only
#6Knowledge47.1Provisional1/4 Measured dimensions
80% interval: 30.8–63.3
broad knowledgePrior only
professional knowledge47.1
supergpqa · super_gpqa
factualityPrior only
retrieval open bookPrior only
#7Multilingual44.2Estimated2/4 Measured dimensions
80% interval: 32.0–56.4
cross language understanding42.6
nova · nova63
multilingual generationPrior only
reasoning transfer45.8
mmlu_prox · mmlu_pro_x
low resource robustnessPrior only
#8Agents38.8RatedGlobal rank #604/4 Measured dimensions
80% interval: 33.7–43.9
planning48.4
deep_planning · gert_labs
tool use29.5
mcp_atlas · mcp_atlas / mcp_tasks · mcp_tasks / tau · tau2_bench / tau · tau3_bench / toolathlon · toolathlon
environment execution42.5
browsecomp · browse_comp / deep_search_qa · deep_search_qa / gdpval_aa · benchlm_agentic_gdpval_aa / terminalbench · benchlm_agentic_terminal_bench2 / wide_research · wide_research
recovery reliability34.9
apex_agents · apex_agents_aa / claw_eval · claw_eval / jobbench · job_bench / researchclaw · research_claw_bench
Specifications
Input and output token limits for this model, plus how it ranks on long-context understanding.
INPUT
262.1Ktokens
≈ 314.6 pages of text
OUTPUT
235.9Ktokens
8K128K1M4M
262.1K
Features
Technical Details
- Input
- Output
- Released
- Jan 2026
- Tokenizer
- Other
- Architecture
- text+image->text
- Moderated
- No
- Supported parameters
- frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Rankings
Excels at
It's decent at
Falls behind in
Detailed scores
Updated: Sep 13, 2026Overall
undefined metric} other undefined metrics}}
Overall score49.0#87 / 112
Overall
undefined metric} other undefined metrics}}
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Pricing
undefined metric} other undefined metrics}}
Input price$0.600/M#96 / 186Output price$3.00/M#99 / 186
Pricing
undefined metric} other undefined metrics}}
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Agents
V3.0undefined metric} other undefined metrics}} · Rated
Score38.8#6080% interval33.7–43.94/4 Measured dimensionsAgentic score28.0#70 / 77Terminal-Bench 2.050.8#44 / 53
Agents
V3.0undefined metric} other undefined metrics}} · Rated
Dimensions and evidence
Planning & decomposition48.4
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
deep_planning · gert_labs · z -0.20 · q 1.00
Tool use29.5
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
mcp_atlas · mcp_atlas · z -3.00 · q 1.00
mcp_tasks · mcp_tasks · z -1.25 · q 0.63
tau · tau2_bench · z 0.74 · q 1.00
tau · tau3_bench · z -1.71 · q 1.00
toolathlon · toolathlon · z -2.21 · q 1.00
Environment & long-horizon execution42.5
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
browsecomp · browse_comp · z -1.48 · q 1.00
deep_search_qa · deep_search_qa · z -0.35 · q 1.00
gdpval_aa · benchlm_agentic_gdpval_aa · z -0.57 · q 1.00
terminalbench · benchlm_agentic_terminal_bench2 · z -0.79 · q 1.00
wide_research · wide_research · z -0.26 · q 1.00
Recovery & completion reliability34.9
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
apex_agents · apex_agents_aa · z -1.17 · q 1.00
claw_eval · claw_eval · z -1.16 · q 1.00
jobbench · job_bench · z -1.88 · q 1.00
researchclaw · research_claw_bench · z -1.37 · q 1.00
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Coding
V3.0undefined metric} other undefined metrics}} · Rated
Score47.3#3080% interval39.6–55.13/4 Measured dimensionsCoding score42.8#69 / 87SWE-bench Verified76.8#27 / 49
Coding
V3.0undefined metric} other undefined metrics}} · Rated
Dimensions and evidence
Code generation50.7
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
livecodebench · live_code_bench_v6 · z -0.32 · q 0.75
scicode · sci_code · z 0.53 · q 1.00
Repository engineering43.7
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
react_native_bench · react_native_evals · z -0.10 · q 1.00
swe_pro · swe_pro · z -1.10 · q 1.00
vibecode · vibe_code_bench · z -0.36 · q 1.00
Debugging & testing47.6
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
swe_multilingual · benchlm_coding_swe_multilingual · z -0.64 · q 1.00
swe_rebench · swe_rebench · z -0.15 · q 1.00
swe_verified · swe_verified · z 0.00 · q 1.00
Tool-assisted development & qualityPrior only
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Reasoning
V3.0undefined metric} other undefined metrics}} · Rated
Score55.5#1780% interval47.2–63.83/4 Measured dimensionsMMLU-Pro87.1%#12 / 129GPQA78.9%#103 / 218
Reasoning
V3.0undefined metric} other undefined metrics}} · Rated
Dimensions and evidence
Abstract logicPrior only
Scientific & causal reasoning56.3
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
critpt · critpt · z 0.00 · q 1.00
gpqa · gpqa · z 0.68 · q 1.00
hle · hle · z 0.60 · q 1.00
Multi-step constraints58.5
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
mmlu_pro · mmlu_pro · z 0.94 · q 1.00
Evidence integration & verification51.8
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
lcr · lcr · z 0.30 · q 1.00
longbench · long_bench_v2 · z 0.00 · q 1.00
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Knowledge
V3.0undefined metric} other undefined metrics}} · Provisional
Score47.180% interval30.8–63.31/4 Measured dimensionsKnowledge score55.5#64 / 83GPQA-D87.6#25 / 32
Knowledge
V3.0undefined metric} other undefined metrics}} · Provisional
Dimensions and evidence
Broad knowledgePrior only
Professional knowledge47.1
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
supergpqa · super_gpqa · z -0.13 · q 1.00
FactualityPrior only
Retrieval & open-book usePrior only
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Artificial Analysis Intelligence Index
LMSpeed rank#72
Score23.5
Sourcebenchlm.ai
UpdatedSep 13, 2026
Confidence4
AA-Omniscience Hallucination Rate
LMSpeed rank#60
Score65.7
Sourcebenchlm.ai
UpdatedSep 13, 2026
Confidence4
Math
V3.0undefined metric} other undefined metrics}} · Estimated
Score49.180% interval40.0–58.13/4 Measured dimensionsMath score62.4#26 / 63AIME 202596.1#1 / 4
Math
V3.0undefined metric} other undefined metrics}} · Estimated
Dimensions and evidence
Foundational mathPrior only
Competition math51.7
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
aime · aime2025 · z 0.63 · q 0.50
aime · aime2026 · z 0.48 · q 1.00
hmmt · hmmt_feb2025 · z 0.31 · q 0.88
hmmt · hmmt_feb2026 · z -0.05 · q 1.00
hmmt · hmmt_nov2025 · z -0.83 · q 1.00
Advanced proofs50.7
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
frontiermath · frontier_math_v2_tier4 · z -0.22 · q 1.00
frontiermath · frontier_math_v2_tiers13 · z 0.12 · q 1.00
Applied & tool-assisted math44.8
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
mmanswer · mm_answer_bench · z -0.65 · q 1.00
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
FrontierMath v2 (Tiers 1-3)V3 evidence
LMSpeed rank#21
Score27.9
Sourcebenchlm.ai
UpdatedSep 13, 2026
Confidence4
FrontierMath v2 (Tier 4)V3 evidence
LMSpeed rank#23
Score4.2
Sourcebenchlm.ai
UpdatedSep 13, 2026
Confidence4
Multilingual
V3.0undefined metric} other undefined metrics}} · Estimated
Score44.280% interval32.0–56.42/4 Measured dimensionsMultilingual score38.2#7 / 11MMLU-ProX82.3#7 / 11
Multilingual
V3.0undefined metric} other undefined metrics}} · Estimated
Dimensions and evidence
Cross-language understanding42.6
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
nova · nova63 · z -1.00 · q 0.88
Multilingual generationPrior only
Reasoning transfer45.8
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
mmlu_prox · mmlu_pro_x · z -0.36 · q 1.00
Low-resource robustnessPrior only
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Multimodal
V3.0undefined metric} other undefined metrics}} · Estimated
Score52.480% interval40.4–64.42/4 Measured dimensionsMultimodal Grounded score65.7#33 / 56MMMU-Pro78.5#16 / 31
Multimodal
V3.0undefined metric} other undefined metrics}} · Estimated
Dimensions and evidence
Perception & OCRPrior only
Document & spatial understandingPrior only
Visual reasoning50.5
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
mmmu_pro · mmmu_pro · z 0.00 · q 1.00
Video & grounded action54.3
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
video_mmmu · video_mmmu · z 1.40 · q 1.00
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Instruction following
V3.0undefined metric} other undefined metrics}} · Estimated
Score54.180% interval42.2–66.02/4 Measured dimensionsInstruction Following score85.8#30 / 52IFEval93.9#5 / 16
Instruction following
V3.0undefined metric} other undefined metrics}} · Estimated
Dimensions and evidence
Constraint following55.6
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
ifeval · ifeval · z 0.54 · q 1.00
Structured outputPrior only
Novel-instruction generalization52.7
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
ifbench · aa_if_bench · z 0.06 · q 1.00
Multi-turn & long instructionsPrior only
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
OpenRouter endpoints
6 endpointsThird-party OpenRouter endpoint data, shown separately from LMSpeed measurements. Some 30-minute live performance fields only appear after syncing with an OpenRouter API key.
| Provider endpoint | Input | Output | 1d uptime | 30m latency | 30m throughput | Context / output |
|---|---|---|---|---|---|---|
Venice venice | $0.532/M | $3.32/M | 100.0% | — | — | undefined tokens / undefined tokens |
SiliconFlow siliconflow/int4 | $0.450/M | $2.25/M | 99.9% | — | — | undefined tokens / undefined tokens |
Amazon Bedrock amazon-bedrock/us-east-2 | $0.600/M | $3/M | 99.7% | — | — | undefined tokens / undefined tokens |
AtlasCloud atlas-cloud/int4 | $0.490/M | $2.50/M | 99.7% | — | — | undefined tokens / undefined tokens |
Phala phala | $0.600/M | $3/M | 99.1% | — | — | undefined tokens / undefined tokens |
Novita novita | $0.570/M | $2.85/M | 97.5% | — | — | undefined tokens / undefined tokens |
Pricing Comparison
Compare Kimi K2.5 API pricing across 182 providers. Prices range from $0.0040/M to $94.90/M. 10dian-API offers the lowest rate at $0.0040/M. 5 providers offer free API credits or a free tier.
| Provider | Health | Model Variant | Group | Input ($/M) | Output ($/M) | Speed (t/s) | First token | Audit |
|---|---|---|---|---|---|---|---|---|
L1 100% | kimi-k2.5 | default | $1.00/request | - | 65.6 t/s | 13.20 s | — | |
L1 100% | kimi-k2.5 | default | $0.643/M Cache read$0.107/M | $3.21/M | 64.9 t/s | 9.95 s | — | |
L1 100% | kimi-k2.5 | default | -54%$0.274/M | -52%$1.44/M | — | — | — | |
L1 100% | kimi-k2.5 | bailian | -52%$0.286/M Cache read$0.050/M | -50%$1.50/M | — | — | — | |
L1 100% | kimi-k2.5 | default | -54%$0.274/M | -52%$1.44/M | — | — | — | |
L1 100% | kimi-k2.5 | default | -9%$0.548/M Cache read$0.055/M | -4%$2.88/M | — | — | — | |
L1 100% | Kimi-K2.5 | Domestic_act_08 | -27%$0.438/M | -23%$2.30/M | — | — | — | |
L1 100% | kimi-k2.5 | default | -34%$0.394/M | -82%$0.532/M | — | — | — | |
L1 100% | kimi-k2.5 | default | -9%$0.548/M Cache read$0.096/M | -4%$2.88/M | — | — | — | |
L1 100% | kimi-k2.5 | default | -93%$0.041/M Cache read$0.0068/M | -93%$0.205/M | — | — | — | |
L1 100% | kimi-k2.5 | Self-Deployed-2 | -85%$0.089/M | -84%$0.467/M | — | — | — | |
L1 100% | kimi-k2.5 | default | -73%$0.164/M | -71%$0.863/M | — | — | — | |
L1 100% | kimi-k2.5 | Model-vip | -73%$0.164/M Cache read$0.016/M | -71%$0.863/M | — | — | — | |
L1 100% | kimi-k2.5 | Kimi-officially | -27%$0.438/M Cache read$0.077/M | -23%$2.30/M | — | — | — | |
L1 100% | kimi-k2.5 | 国产模型 | -15%$0.509/M | -11%$2.67/M | — | — | — | |
L1 100% | kimi-k2.5 | default | $4.00/M | $21.00/M | — | — | — | |
DeadlySignal API Free | L1 99% L2 100% | kimi-k2.5 | default | Free | Free | — | — | — |
L1 100% | Kimi-K2.5 | default | -32%$0.405/M | -32%$2.03/M | — | — | — | |
L1 100% | kimi-k2.5 | default | -32%$0.405/M | -32%$2.03/M | — | — | — | |
L1 100% L2 0% | kimi-k2.5 | default | -4%$0.574/M Cache read$0.077/M | -20%$2.41/M | — | — | — |
Showing 20 model IDs of 80.
Alternatives & Similar Models
MiniMax M2.5
minimax-m2-5
MiniMax M2.5 is MiniMax's flagship text model for coding and agents, with SOTA-level programming and agentic performance, improved token efficiency, and fast high-TPS API deployment.
214 shared providers
DeepSeek V4 Flash
deepseek-v4-flash
DeepSeek V4 Flash is a fast, cost-efficient language model in the DeepSeek V4 family, optimized for low-latency chat, coding assistance, and high-throughput API workloads while retaining strong reasoning quality.
211 shared providers
GLM-5.1
glm-5-1
Zhipu GLM-5.1 is a next-generation GLM model aimed at frontier reasoning, coding, and bilingual agent applications.
211 shared providers
GLM-5
glm-5
Zhipu GLM-5 is Zhipu flagship GLM series model with enhanced reasoning, agent capabilities, and strong performance on Chinese enterprise and coding scenarios.
209 shared providers
DeepSeek V4 Pro
deepseek-v4-pro
DeepSeek V4 Pro is the professional-tier DeepSeek V4 model, targeting frontier reasoning, coding, and agent workflows with maximum capability.
207 shared providers
GPT-5.4
gpt-5-4
OpenAI GPT-5.4 extends the GPT-5 family with stronger instruction following, deeper tool use, and improved performance on coding, math, and long-document analysis.
205 shared providers
Frequently Asked Questions
- What benchmark data does Kimi K2.5 include?
- LMSpeed shows Kimi K2.5 benchmark context, API price, output speed, first-token latency, and provider data across 187 providers when those signals are available.
- What is the Kimi K2.5 API price?
- Kimi K2.5 has pricing from undefined provider} other undefined providers}}, ranging from $0.0040/M to $94.90/M. 10dian-API has the lowest listed price.
- What does the Kimi K2.5 API pricing table include?
- The Kimi K2.5 API pricing table compares 187 providers by input price, output price, free tier status, speed, first-token latency, and recent health data when available.
- Which provider has the cheapest Kimi K2.5 API pricing?
- 10dian-API currently has the lowest listed Kimi K2.5 price at $0.0040/M across undefined provider} other undefined providers}}.
- Can I compare Kimi K2.5 API price and speed together?
- Yes. LMSpeed shows Kimi K2.5 API price, output speed, first-token latency, and provider health on the same page so you can compare cost and performance together.
- Is Kimi K2.5 API free?
- Yes, Kimi K2.5 free API options are available through 5 providerundefined other undefined} on LMSpeed, including 兔子API, Zero API, 兔子API, 兔子API, DeadlySignal API. These providers offer free API credits or a free tier with no per-token charges.
- Where can I get Kimi K2.5 free API access?
- LMSpeed currently lists 5 free API providerundefined other undefined} for Kimi K2.5: 兔子API, Zero API, 兔子API, 兔子API, DeadlySignal API. Check each provider row before using it because free tier limits can change.