DeepSeek
·Released on Aug 12, 2026DeepSeek V4 Pro 0813 API Benchmarks, Pricing & Provider Data
Compare DeepSeek V4 Pro 0813 with another model
Choose a model to open its comparison page.
LLM
DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts model from DeepSeek. This is the GA release of DeepSeek V4 Pro.
Quality
#26of 100
67.0
LMSpeed score
Category Performance
Observed-capability estimates with uncertainty reported separately from independent benchmark families.
- Coverage
- 6 / 8
- Methodology
- V3.0
#1Reasoning61RatedGlobal rank #23/4 Measured dimensions
80% interval: 52.6–69.3
abstract logicPrior only
scientific causal62.7
critpt · critpt / gpqa · aa_gpqa_diamond / hle · aa_hle
multistep constraints59
mmlu_pro · mmlu_pro
evidence verification61.1
lcr · lcr / mrcr · mrcr1m
#2Math58.9Estimated2/4 Measured dimensions
80% interval: 46.8–71.1
foundational mathPrior only
competition math61.7
hmmt · hmmt_feb2026
proof frontier56.2
imo · imo_answer_bench
applied tool mathPrior only
#3Knowledge58.4Provisional1/4 Measured dimensions
80% interval: 40.4–76.5
broad knowledgePrior only
professional knowledgePrior only
factuality58.4
simpleqa · simple_qa
retrieval open bookPrior only
#4Agents57.5RatedGlobal rank #133/4 Measured dimensions
80% interval: 49.8–65.2
planningPrior only
tool use58.1
mcp_atlas · mcp_atlas / tau · aa_tau3_banking / tau · tau2_bench / toolathlon · toolathlon
environment execution55.9
browsecomp · browse_comp / gdpval_aa · benchlm_agentic_gdpval_aa / terminalbench · benchlm_agentic_terminal_bench2
recovery reliability58.6
apex_agents · apex_agents_aa / cyber_gym · cyber_gym
#5Coding57.1RatedGlobal rank #104/4 Measured dimensions
80% interval: 50.9–63.2
code generation58.5
scicode · aa_sci_code
repository engineering62.4
nl2repo · nl2_repo / swe_pro · swe_pro / vibecode · vibe_code_bench
debugging testing56.9
swe_multilingual · benchlm_coding_swe_multilingual / swe_verified · swe_verified
tooling quality50.5
terminalbench · aa_terminal_bench21 / terminalbench · benchlm_coding_terminal_bench2
#6Instruction following54.9Provisional1/4 Measured dimensions
80% interval: 38.9–70.9
constraint followingPrior only
structured outputPrior only
novel instruction generalization54.9
ifbench · aa_if_bench
long multiturn instructionPrior only
No data:MultilingualMultimodal
Specifications
Input and output token limits for this model, plus how it ranks on long-context understanding.
INPUT
1.0Mtokens
≈ 1.3K pages of text
OUTPUT
384Ktokens
8K128K1M4M
1.0M
Features
Technical Details
- Input
- Output
- Released
- Aug 2026
- Tokenizer
- DeepSeek
- Architecture
- text->text
- Moderated
- No
- Supported parameters
- frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Rankings
Excels at
It's decent at
Falls behind in
Detailed scores
Updated: Aug 23, 2026Overall
undefined metric} other undefined metrics}}
Overall score67.0#26 / 100DeepSWE62.7#8 / 16
Overall
undefined metric} other undefined metrics}}
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Agents
V3.0undefined metric} other undefined metrics}} · Rated
Score57.5#1380% interval49.8–65.23/4 Measured dimensionsAgentic score68.2#25 / 69Terminal-Bench 2.067.9#20 / 53
Agents
V3.0undefined metric} other undefined metrics}} · Rated
Dimensions and evidence
Planning & decompositionPrior only
Tool use58.1
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
mcp_atlas · mcp_atlas · z 0.04 · q 1.00
tau · aa_tau3_banking · z 0.59 · q 1.00
tau · tau2_bench · z 0.77 · q 1.00
toolathlon · toolathlon · z 0.38 · q 1.00
Environment & long-horizon execution55.9
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
browsecomp · browse_comp · z 0.28 · q 1.00
gdpval_aa · benchlm_agentic_gdpval_aa · z 0.40 · q 1.00
terminalbench · benchlm_agentic_terminal_bench2 · z 0.27 · q 1.00
Recovery & completion reliability58.6
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
apex_agents · apex_agents_aa · z -0.04 · q 1.00
cyber_gym · cyber_gym · z 0.72 · q 1.00
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Coding
V3.0undefined metric} other undefined metrics}} · Rated
Score57.1#1080% interval50.9–63.24/4 Measured dimensionsCoding score57.6#36 / 81Codeforces3206.0#1 / 4
Coding
V3.0undefined metric} other undefined metrics}} · Rated
Dimensions and evidence
Code generation58.5
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
scicode · aa_sci_code · z 0.99 · q 1.00
Repository engineering62.4
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
nl2repo · nl2_repo · z 3.00 · q 1.00
swe_pro · swe_pro · z -0.26 · q 1.00
vibecode · vibe_code_bench · z 0.86 · q 1.00
Debugging & testing56.9
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
swe_multilingual · benchlm_coding_swe_multilingual · z 0.30 · q 1.00
swe_verified · swe_verified · z 0.80 · q 1.00
Tool-assisted development & quality50.5
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
terminalbench · aa_terminal_bench21 · z -0.10 · q 1.00
terminalbench · benchlm_coding_terminal_bench2 · z 0.00 · q 1.00
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
AA Terminal-Bench 2.1V3 evidence
LMSpeed rank#10
Score78.7
Sourcebenchlm.ai
UpdatedAug 23, 2026
Confidence2
Reasoning
V3.0undefined metric} other undefined metrics}} · Rated
Score61#280% interval52.6–69.33/4 Measured dimensionsMMLU-Pro87.5%#9 / 129GPQA90.1%#28 / 211
Reasoning
V3.0undefined metric} other undefined metrics}} · Rated
Dimensions and evidence
Abstract logicPrior only
Scientific & causal reasoning62.7
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
critpt · critpt · z 0.83 · q 1.00
gpqa · aa_gpqa_diamond · z 1.27 · q 1.00
hle · aa_hle · z 0.95 · q 1.00
Multi-step constraints59
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
mmlu_pro · mmlu_pro · z 1.00 · q 1.00
Evidence integration & verification61.1
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
lcr · lcr · z 0.36 · q 1.00
mrcr · mrcr1m · z 1.52 · q 0.75
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Knowledge
V3.0undefined metric} other undefined metrics}} · Provisional
Score58.480% interval40.4–76.51/4 Measured dimensionsKnowledge score77.0#20 / 69SimpleQA57.9#1 / 4
Knowledge
V3.0undefined metric} other undefined metrics}} · Provisional
Dimensions and evidence
Broad knowledgePrior only
Professional knowledgePrior only
Factuality58.4
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
simpleqa · simple_qa · z 1.39 · q 0.17
Retrieval & open-book usePrior only
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Artificial Analysis Intelligence Index
LMSpeed rank#15
Score53.2
Sourcebenchlm.ai
UpdatedAug 23, 2026
Confidence2
AA-Omniscience Hallucination Rate
LMSpeed rank#2
Score94.1
Sourcebenchlm.ai
UpdatedAug 23, 2026
Confidence2
Math
V3.0undefined metric} other undefined metrics}} · Estimated
Score58.980% interval46.8–71.12/4 Measured dimensionsHMMT Feb 202695.2#2 / 18IMOAnswerBench89.8#2 / 7
Math
V3.0undefined metric} other undefined metrics}} · Estimated
Dimensions and evidence
Foundational mathPrior only
Competition math61.7
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
hmmt · hmmt_feb2026 · z 1.69 · q 1.00
Advanced proofs56.2
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
imo · imo_answer_bench · z 0.34 · q 0.88
Applied & tool-assisted mathPrior only
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Multilingual
V3.0undefined metric} other undefined metrics}} · No data
No data80% interval30.8–69.20/4 Measured dimensions
Multilingual
V3.0undefined metric} other undefined metrics}} · No data
Dimensions and evidence
Cross-language understandingPrior only
Multilingual generationPrior only
Reasoning transferPrior only
Low-resource robustnessPrior only
Multimodal
V3.0undefined metric} other undefined metrics}} · No data
80% interval30.8–69.20/4 Measured dimensionsDesign Arena Website1258.0#33 / 75
Multimodal
V3.0undefined metric} other undefined metrics}} · No data
Dimensions and evidence
Perception & OCRPrior only
Document & spatial understandingPrior only
Visual reasoningPrior only
Video & grounded actionPrior only
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
Instruction following
V3.0undefined metric} other undefined metrics}} · Provisional
Score54.980% interval38.9–70.91/4 Measured dimensionsAA-IFBench76.5#10 / 84
Instruction following
V3.0undefined metric} other undefined metrics}} · Provisional
Dimensions and evidence
Constraint followingPrior only
Structured outputPrior only
Novel-instruction generalization54.9
undefined family} other undefined families}} · undefined metric} other undefined metrics}}
ifbench · aa_if_bench · z 0.38 · q 1.00
Multi-turn & long instructionsPrior only
Benchmark
LMSpeed rank
Score
Source
Updated
Confidence
OpenRouter endpoints
14 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 |
|---|---|---|---|---|---|---|
Alibaba alibaba | $1.16/M | $3.48/M | 100.0% | — | — | undefined tokens / undefined tokens |
DeepSeek deepseek | $0.660/M | $1.98/M | 100.0% | — | — | undefined tokens / undefined tokens |
DigitalOcean digitalocean | $1.32/M | $3.96/M | 99.9% | — | — | undefined tokens / — |
Phala phala | $1.45/M | $4.36/M | 99.9% | — | — | undefined tokens / undefined tokens |
Fireworks fireworks | $1.32/M | $3.96/M | 99.9% | — | — | undefined tokens / — |
Novita novita/fp8 | $1.32/M | $3.96/M | 99.9% | — | — | undefined tokens / undefined tokens |
GMICloud gmicloud/fp8 | $1.12/M | $3.37/M | 99.9% | — | — | undefined tokens / — |
Sail Research sail-research/fp4 | $1.32/M | $3.96/M | 99.8% | — | — | undefined tokens / undefined tokens |
BaseTen baseten/fp4 | $1.32/M | $3.96/M | 99.5% | — | — | undefined tokens / undefined tokens |
StreamLake streamlake | $1.32/M | $3.96/M | 99.4% | — | — | undefined tokens / undefined tokens |
Parasail parasail/fp8 | $1.32/M | $3.96/M | 99.4% | — | — | undefined tokens / undefined tokens |
SiliconFlow siliconflow/fp8 | $1.32/M | $3.96/M | 98.6% | — | — | undefined tokens / undefined tokens |
Together together | $1.32/M | $3.96/M | 98.4% | — | — | undefined tokens / — |
Cloudflare cloudflare | $1.32/M | $3.96/M | 98.3% | — | — | undefined tokens / undefined tokens |