The Model Index
Up to four models

DeepSeek-V4-Pro vs Step 5 Preview

On the index (GPT-4 = 100), DeepSeek-V4-Pro scores 163 and Step 5 Preview 163. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.

Next to DeepSeek-V4-Pro
Presets
DeepSeek-V4-Pro
163147–178
WeightsOpen
ReleasedAug 2026
Price in / out$0.11 / $4.44
Context1.05M
Core tests taken6 of 9
Step 5 Preview
163145–180
WeightsClosed
ReleasedSep 2026
Price in / out$1.0 / $2.7
Context1M
Core tests taken2 of 9

DeepSeek-V4-Pro and Step 5 Preview are too close to call (0 points apart, within the uncertainty), ahead on 0 of 2 shared core tests, at a similar price.

Test by testthird-party results

DeepSeek-V4-ProStep 5 Preview
0%25%50%75%100%MATHSProofBench v1.1ProofBench v1.1CODINGTerminal-Bench 4.0Terminal-Bench 4.0Code MigrationCode MigrationIOIIOIVibe Code Bench v1.1Vibe Code Bench v1.1AGENTSSRE BenchSRE BenchTerminal-Bench-ScienceTerminal-Bench-SciencePROFESSIONAL WORKFinance Agent (v2)Finance Agent (v2)EMBEMBHLAB-Vals-1.1HLAB-Vals-1.1Legal Research BenchLegal Research BenchTax Agent BenchTax Agent Bench
Every result as a table
TestDeepSeek-V4-ProStep 5 Preview
Maths
ProofBench v1.150.0%42.0%
Coding
Terminal-Bench 4.014.1%31.8%
Code Migration41.5%39.0%
IOI51.6%38.1%
Vibe Code Bench v1.182.3%70.4%
Agents
SRE Bench1.9%0.4%
Terminal-Bench-Science4.3%2.9%
Professional work
Finance Agent (v2)50.4%50.7%
EMB52.8%57.1%
HLAB-Vals-1.110.8%12.9%
Legal Research Bench40.9%40.4%
Tax Agent Bench26.8%27.3%

Tests at least two of these models have taken. Bold rows are the core tests.

This page as a chartDownload PNG