Agentic
Like-for-like- DeepSeek V4 Pro
- 59.1
- MAI-Thinking-1
- 46.0
- Weighted basis
- 1 vs 1 rows
- Reading
- DeepSeek V4 Pro leads
Model comparison
DeepSeek V4 Pro vs MAI-Thinking-1
Updated August 7, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
DeepSeek V4 Pro has the higher public score estimate, 60.01 versus 51.03, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
9 results are shared. Category rows based on different benchmark sets are marked directional and do not name a winner.
Which one for your work
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.
Coding work
Code generation, repair, and software-engineering tasks
MAI-Thinking-1
MAI-Thinking-1 leads on the same 2 weighted benchmark rows.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
DeepSeek V4 Pro
DeepSeek V4 Pro leads on the same 1 weighted benchmark row.
Confidence: limited
Long documents
Prompts that approach the documented context limit
DeepSeek V4 Pro
DeepSeek V4 Pro has the larger documented context window.
Confidence: documented
Show secondary and unsupported calls
Chat turn cost
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Repository review cost
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
14 DeepSeek V4 Pro only
9 Shared
5 MAI-Thinking-1 only
- Shared results
- 9
- DeepSeek V4 Pro only
- 14
- MAI-Thinking-1 only
- 5
- Like-for-like categories
- 2 / 8
2 categories use different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Category results, on a stated basis
Each row states whether both averages use the same weighted benchmark set. Directional and not-comparable rows remain visible, but they never receive a winner in this template.
Coding
Like-for-like- DeepSeek V4 Pro
- 65.3
- MAI-Thinking-1
- 65.5
- Weighted basis
- 2 vs 2 rows
- Reading
- MAI-Thinking-1 leads
Knowledge
Directional only- DeepSeek V4 Pro
- 41.3
- MAI-Thinking-1
- 72.5
- Weighted basis
- 4 vs 3 rows
- Reading
- Directional only
Math
Directional only- DeepSeek V4 Pro
- 31.7
- MAI-Thinking-1
- 89.7
- Weighted basis
- 1 vs 2 rows
- Reading
- Directional only
Reasoning
Not comparable- DeepSeek V4 Pro
- Not measured
- MAI-Thinking-1
- Not measured
- Weighted basis
- 0 vs 0 rows
- Reading
- Not comparable
Multilingual
Not comparable- DeepSeek V4 Pro
- Not measured
- MAI-Thinking-1
- Not measured
- Weighted basis
- 0 vs 0 rows
- Reading
- Not comparable
Multimodal
Not comparable- DeepSeek V4 Pro
- Not measured
- MAI-Thinking-1
- Not measured
- Weighted basis
- 0 vs 0 rows
- Reading
- Not comparable
Instruction following
Not comparable- DeepSeek V4 Pro
- Not measured
- MAI-Thinking-1
- 85.0
- Weighted basis
- 0 vs 1 rows
- Reading
- Not comparable
| Category | DeepSeek V4 Pro | MAI-Thinking-1 | Weighted basis | Reading |
|---|---|---|---|---|
| Agentic | 59.1 | 46.0 | Like-for-like1 vs 1 rows | DeepSeek V4 Pro leads |
| Coding | 65.3 | 65.5 | Like-for-like2 vs 2 rows | MAI-Thinking-1 leads |
| Knowledge | 41.3 | 72.5 | Directional only4 vs 3 rows | Directional only |
| Math | 31.7 | 89.7 | Directional only1 vs 2 rows | Directional only |
| Reasoning | Not measured | Not measured | Not comparable0 vs 0 rows | Not comparable |
| Multilingual | Not measured | Not measured | Not comparable0 vs 0 rows | Not comparable |
| Multimodal | Not measured | Not measured | Not comparable0 vs 0 rows | Not comparable |
| Instruction following | Not measured | 85.0 | Not comparable0 vs 1 rows | Not comparable |
Shape of the matched evidence
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.
HMMT Feb 2026
Math
DeepSeek V4 Pro: 31.7%MAI-Thinking-1: 84.9%Normalized gap 53.2DeepSeek V4 Pro source MAI-Thinking-1 sourceSimpleQA
Knowledge
DeepSeek V4 Pro: 45%MAI-Thinking-1: 31%Normalized gap 14.0DeepSeek V4 Pro source MAI-Thinking-1 sourceTerminal-Bench 2.0
Agentic
DeepSeek V4 Pro: 59.1%MAI-Thinking-1: 46%Normalized gap 13.1DeepSeek V4 Pro source MAI-Thinking-1 sourceGPQA
Knowledge
DeepSeek V4 Pro: 72.9%MAI-Thinking-1: 84.2%Normalized gap 11.3DeepSeek V4 Pro source MAI-Thinking-1 sourceMMLU-Pro
Knowledge
DeepSeek V4 Pro: 82.9%MAI-Thinking-1: 85%Normalized gap 2.1DeepSeek V4 Pro source MAI-Thinking-1 source
What each workload costs
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.
Chat turn
1K fresh input + 500 output tokens
- DeepSeek V4 Pro
- $0.00087
- Fits in one request
- MAI-Thinking-1
- API rate not published
- Fits in one request
MAI-Thinking-1 has no comparable published API token rate.
Repository review
50K fresh input + 3K output tokens
- DeepSeek V4 Pro
- $0.02436
- Fits in one request
- MAI-Thinking-1
- API rate not published
- Fits in one request
MAI-Thinking-1 has no comparable published API token rate.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- DeepSeek V4 Pro
- $0.01812
- Fits in one request
- MAI-Thinking-1
- API rate not published
- Fits in one request
- Cached-input rate unavailable
MAI-Thinking-1 has no comparable published API token rate.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationDeepSeek V4 ProMAI-Thinking-1
Context window
Maximum documented context; output-token limits may be lower.
DeepSeek V4 Pro
MAI-Thinking-1
256K
API model ID
DeepSeek V4 Pro
deepseek-v4-pro
DeepSeek models and pricingMAI-Thinking-1
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
DeepSeek V4 Pro
$0.003625 per 1M cached input tokens
MAI-Thinking-1
No comparable hosted API rate
Documented inputs
DeepSeek V4 Pro
MAI-Thinking-1
Not sourced
Documented outputs
DeepSeek V4 Pro
MAI-Thinking-1
Not sourced
Provider availability
DeepSeek V4 Pro
Preview · DeepSeek API, open weights
DeepSeek V4 Flash 0731 updateMAI-Thinking-1
Not sourced
Reasoning profile
DeepSeek V4 Pro
Non-Reasoning
MAI-Thinking-1
Reasoning
Weight access
DeepSeek V4 Pro
Open Weight
MAI-Thinking-1
Proprietary
License
DeepSeek V4 Pro
Open Weight
MAI-Thinking-1
Proprietary
Release date
DeepSeek V4 Pro
2026-04-24
MAI-Thinking-1
2026-06-02
If you already use one of these models
- Deployment change
- The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
- Quality signal
- DeepSeek V4 Pro has the higher public score estimate, 60.01 versus 51.03, but the 90% score intervals overlap.
- Workload cost
- A complete comparable API-rate estimate is not available for both models.
- Context tradeoff
- DeepSeek V4 Pro has the larger documented window (1M).
Run the same representative tasks against both endpoints before changing production traffic.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence28 rows
Agentic
Terminal-Bench 2.0
DeepSeek V4 Pro59.1%SourceMAI-Thinking-146%SourceDeepSeek V4 Pro leads this result
MCP Atlas
DeepSeek V4 Pro69.4%SourceMAI-Thinking-1—Not directly comparable
Toolathlon
DeepSeek V4 Pro46.3%SourceMAI-Thinking-1—Not directly comparable
Claw-Eval
DeepSeek V4 Pro59.8%SourceMAI-Thinking-1—Not directly comparable
Gert Labs
DeepSeek V4 Pro50.28%SourceMAI-Thinking-1—Not directly comparable
ResearchClawBench
DeepSeek V4 Pro17.1%SourceMAI-Thinking-1—Not directly comparable
Coding
LiveCodeBench Pass@1-COT
DeepSeek V4 Pro56.8%SourceMAI-Thinking-1—Not directly comparable
SWE-bench Verified
DeepSeek V4 Pro73.6%SourceMAI-Thinking-173.5%SourceDeepSeek V4 Pro leads this result
SWE-bench Pro
DeepSeek V4 Pro52.1%SourceMAI-Thinking-152.8%SourceMAI-Thinking-1 leads this result
SWE Multilingual
DeepSeek V4 Pro69.8%SourceMAI-Thinking-1—Not directly comparable
Terminal-Bench 2.0
DeepSeek V4 Pro59.1%SourceMAI-Thinking-146.0%SourceDeepSeek V4 Pro leads this result
LiveCodeBench v6
DeepSeek V4 Pro—MAI-Thinking-187.7%SourceNot directly comparable
Reasoning
Knowledge
MMLU-Pro
DeepSeek V4 Pro82.9%SourceMAI-Thinking-185%SourceMAI-Thinking-1 leads this result
SimpleQA
DeepSeek V4 Pro45%SourceMAI-Thinking-131%SourceDeepSeek V4 Pro leads this result
Chinese-SimpleQA
DeepSeek V4 Pro75.8%SourceMAI-Thinking-1—Not directly comparable
GPQA
DeepSeek V4 Pro72.9%SourceMAI-Thinking-184.2%SourceMAI-Thinking-1 leads this result
GPQA-D
DeepSeek V4 Pro72.9%SourceMAI-Thinking-184.2%SourceMAI-Thinking-1 leads this result
HLE
DeepSeek V4 Pro7.7%SourceMAI-Thinking-1—Not directly comparable
Math
HMMT Feb 2026
DeepSeek V4 Pro31.7%SourceMAI-Thinking-184.9%SourceMAI-Thinking-1 leads this result
IMOAnswerBench
DeepSeek V4 Pro35.3%SourceMAI-Thinking-1—Not directly comparable
Apex
DeepSeek V4 Pro0.4%SourceMAI-Thinking-1—Not directly comparable
Apex Shortlist
DeepSeek V4 Pro9.2%SourceMAI-Thinking-1—Not directly comparable
AIME 2025
DeepSeek V4 Pro—MAI-Thinking-197%SourceNot directly comparable
AIME26
DeepSeek V4 Pro—MAI-Thinking-194.5%SourceNot directly comparable
Instruction following
IFBench
DeepSeek V4 Pro—MAI-Thinking-185%SourceNot directly comparable
Frequently asked questions
Which is better, DeepSeek V4 Pro or MAI-Thinking-1?
DeepSeek V4 Pro has the higher public score estimate, 60.01 versus 51.03, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, DeepSeek V4 Pro or MAI-Thinking-1?
MAI-Thinking-1 leads the like-for-like coding comparison across 2 shared weighted benchmark rows.
Which is better for agentic tasks, DeepSeek V4 Pro or MAI-Thinking-1?
DeepSeek V4 Pro leads the like-for-like agentic tasks comparison across 1 shared weighted benchmark row.
Which costs less, DeepSeek V4 Pro or MAI-Thinking-1?
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Which has the larger context window, DeepSeek V4 Pro or MAI-Thinking-1?
DeepSeek V4 Pro has the larger documented context window: 1M, compared with 256K.
Related comparisons
Last updated August 7, 2026
Watch DeepSeek V4 Pro vs MAI-Thinking-1
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.