DeepSeek V4 Pro 0424 vs DeepSeek V4 Pro 0813
Different efforts: high · non-thinking · Compare all at high
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
DeepSeek V4 Pro 042435.2
DeepSeek V4 Pro 081310.6
Terminal-Bench 4.0 Terminal-Bench 4.0 · source setting 2: Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite
DeepSeek V4 Pro 0424 · max14.6
DeepSeek V4 Pro 08131.5
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
DeepSeek V4 Pro 0424 · max56.3
DeepSeek V4 Pro 081334.9
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
DeepSeek V4 Pro 0424—
DeepSeek V4 Pro 0813 · high1,446
| Benchmark | DeepSeek V4 Pro 0424 | DeepSeek V4 Pro 0813 |
|---|---|---|
| Humanity's Last Exam | 35.2 | 10.6 |
| Terminal-Bench 4.0 | 14.6 · max | 1.5 |
| AutomationBench-AA | 56.3 · max | 34.9 |
| Arena Text · creative writing | — | 1,446 · high |
Specs
| Spec | DeepSeek V4 Pro 0424 | DeepSeek V4 Pro 0813 |
|---|---|---|
| Input, per 1M | !$0.43 | !$1.32 |
| Output, per 1M | !$0.87 | !$3.96 |
| Cost per task? | — | $0.36 |
| Speed, tok/s | 102 | 137 |
| Context | 1M | 1M |
| Weights | Open | Open |
| Released | Apr 2026 | Aug 2026 |
| Memory (4-bit) | — | 990 GB |
| Licence | — | MIT |
| Commercial use | — | Allowed |
| See images | ? | — |
| Read PDFs | ? | — |
| Hear audio | ? | — |
| Watch video | ? | — |
| Call tools | ? | ✓ |
| Force a tool call | ? | ✓ |
| Return strict JSON | ? | ✓ |
| Search the web | ? | ? |
| Run code | ? | ? |
| Use a computer | ? | ? |
What breaks when you switch
DeepSeek V4 Pro 0424 → DeepSeek V4 Pro 0813
Pass back reasoning_content · Chat Completions›
With thinking and tools, return all prior reasoning_content even on turns without tool calls or the API returns HTTP 400.
Disable thinking mode first · Chat Completions›
Required and named tool choices return HTTP 400 in thinking mode; disable thinking first.
Parameters have no effect›
In thinking mode, temperature, presence_penalty and frequency_penalty have no effect.
top_p value is ignored›
In thinking mode, top_p is fixed at 0.95 with tools and 1.0 otherwise; in non-thinking mode it takes effect.
Stateless API · Responses›
Responses does not support previous_response_id or conversation; preserve conversation history in your client.
DeepSeek V4 Pro 0813 → DeepSeek V4 Pro 0424