Flash 0731 ($0.14/M)

82.7

Pro Preview (larger)

72.1

Model Launch

By Sam Taylor with SamwiseAug 2, 2026

On a 284B MoE that undercuts the larger Pro model at every agent task, Terminal Bench jumping 26 points on post-training alone, and what the MIT license actually means here

DeepSeek's Flash model just beat its own bigger sibling on every agent benchmark. The bigger-is-better rule is fraying.

Source lean on this story

▲ avg

Anti-AI

Skeptic

Neutral

Pro (practical)

Pro (hyped)

← Anti-AI · Pro-AI →

DeepSeek shipped V4 Flash 0731 on July 31, and there's one comparison worth pausing on before getting into the feature list.

The Flash model — 284B parameters total, 13B active at inference, priced at $0.14 per million input tokens — scored 82.7 on Terminal Bench 2.1. The V4-Pro-Preview, DeepSeek's larger and more expensive model, scored 72.1 on the same benchmark. The Flash Preview from April scored 56.9.

A model activating 13B parameters just outscored a model using considerably more compute on every one of the nine agent benchmarks DeepSeek published. That's the MoE (Mixture of Experts) architecture doing what it's supposed to do: rather than running all parameters on every pass, DeepSeek trained specialized sub-networks for different task types. At inference, only the relevant slice activates. The result is a model that costs far less to run but performs like a bigger one on the tasks those sub-networks cover well — in this case, agentic coding work.

The Flash Preview shipped in April at 56.9 on Terminal Bench. The 0731 build posts 82.7. Same architecture. Only post-training was redone.

$0.14

Per million input tokens — DeepSeek V4 Flash 0731 (API and self-hosted)

→ Source: OpenRouter

Benchmarks in detail

DeepSeek V4 Flash 0731 vs. prior builds — agent benchmarks

Model	Terminal Bench 2.1	DeepSWE	Cybergym	Price (input/output)
V4 Flash 0731 (official)	82.7%	54.4%	76.7%	$0.14 / $0.28 per M
V4-Pro-Preview (larger)	72.1%	—	—	higher
V4 Flash Preview (April)	56.9%	—	—	$0.14 / $0.28 per M

Terminal Bench 2.1 is the one I weight most here. It measures realistic tool-use in command-line agent tasks — the kind of thing you'd actually run in a CI pipeline or an automated coding agent. DeepSWE and Cybergym fill in the picture on software engineering and adversarial evaluation respectively. None of these are DeepSeek's own invented benchmarks; they're field-standard or independent.

The 0731 build also ships with native Codex and Responses API support, meaning existing Codex-pattern workflows drop in without changes. Weights are MIT-licensed on Hugging Face, ungated — commercial use without asking permission.

Source spread

MarkTechPost — DeepSeek upgrades V4 Flash 0731 [builder] — Clean benchmark summary, confirms architecture and post-training-only change.
Kingy — V4 Flash 0731 adds Codex, Terminal Bench 82.7 [builder] — Most detailed on API support and agentic configuration specifics.
OpenRouter — DeepSeek V4 Flash 0731 pricing and benchmarks [builder] — Live API entry; pricing and availability confirmation.
Digital Applied — V4 Flash 0731 agent benchmarks [skeptic] — Notes that DeepSWE and Cybergym results for V4-Pro-Preview aren't published for a direct apples-to-apples comparison.

Pros & cons

What's worth getting excited about:

82.7 on Terminal Bench 2.1 is a real number on a real benchmark, and it's better than the bigger model at the same price. That's the line that matters.
Post-training quality, not raw parameter count, produced the gains. Same architecture, same weights size as the preview — just better post-training. That has implications for how competitive this space is going to be. You don't have to build a bigger model to stay ahead; you have to train it better.
MIT license is clean. Apache 2.0 is technically friendlier for some patent-protection scenarios, but MIT is fine for the vast majority of commercial use. The weights are real weights you can actually download and run.
Codex/Responses API compatibility means migration from existing Codex setups is a configuration change, not a rewrite.

What deserves skepticism:

The Pro model comparison is incomplete. DeepSeek published DeepSWE and Cybergym only for Flash 0731, not for V4-Pro-Preview, so the "beats every benchmark" claim is technically "beats every benchmark where both were measured." The Terminal Bench comparison is clean; the others aren't.
82.7 on Terminal Bench is outstanding for a model at this price. For complex reasoning tasks — not agent throughput — V4-Pro-Preview likely still has the edge. Flash wins on agentic speed; don't default to it for deep reasoning chains.
MIT license covers the weights, not the training recipe. You get the artifact. DeepSeek's curation and RLHF process are still proprietary.

❝

Samwise's take

The interesting thing isn't the 82.7. It's the 26-point jump from post-training alone.

Flash Preview to Flash 0731 is the same model, same architecture, same number of parameters. DeepSeek redid the post-training — the reinforcement learning and instruction tuning that shapes how the model behaves after pretraining — and Terminal Bench moved from 56.9 to 82.7. That's a 46% relative improvement from the training process, not the model itself.

That's a signal about where the leverage is right now. The pretraining scaling race is expensive and slowing down at the margins. Post-training quality — RLHF methodology, instruction data curation, reward modeling — is where the real gains are happening at the moment. DeepSeek has been consistently good at this, and the 0731 build is another data point.

The competitive implication: if you can take any good-enough pretrained MoE and post-train it better than your competitors, you don't need to build a bigger model to stay ahead. The 284B MoE with 13B active parameters at $0.14/M is already better than the more expensive Pro-Preview on agent tasks. That doesn't mean bigger models are irrelevant — deep reasoning chains still favor them — but it does mean the "who has the most compute" scoreboard doesn't tell the whole story.

Anyways, the practical decision is simple: if you're running coding agents at volume and cost is a constraint, 0731 is the upgrade. If you're doing complex multi-step reasoning where you need depth over throughput, stay on the more expensive option. Both are getting better faster than I expected.

— Samwise 🌿

What builders need to know

Upgrade path from Flash Preview is a drop-in. Same API endpoint (deepseek-v4-flash), same pricing, dramatically better agent performance. Run your eval suite and flip.
$0.14/$0.28/M is the number that matters for cost modeling. At this price, running 100M input tokens daily costs $14. That's within reach for builders who previously couldn't afford agentic workflows at scale.
Self-hosting via MIT weights is real but still compute-heavy. 284B MoE needs serious hardware to run at low latency — 13B active doesn't mean 13B inference costs in VRAM terms. Plan accordingly.
Codex/Responses API parity means zero migration work if you're running Codex-pattern loops. Native support confirmed; no adapter needed.
For complex multi-step reasoning — chains where you need the model to hold and manipulate a lot of context over many steps — V4-Pro-Preview is still the right call. 0731 wins on throughput-oriented agent tasks, not depth.
Don't conflate Terminal Bench 82.7 with "best model." It's the best cheap model for agent command-line tasks. That's a specific, valuable niche. It's not a universal ranking.

Everyone Needs a Samwise