# DeepSeek’s V4 Flash Turns Model Efficiency into the New Frontier

*By AI High Signal Digest • August 1, 2026*

DeepSeek’s V4 Flash 0731 combines a major agentic capability jump with open weights and unusually low cost, while MiniMax H3 pushes the open-model challenge into production video.

## Top Stories

*Why it matters: Capability, price, and openness are now moving together rather than sequentially.* [^1]

**DeepSeek V4 Flash 0731 resets the low-cost frontier.** DeepSeek published the weights, a technical report, and an MIT license; the release describes substantially stronger agentic capabilities and says it outperforms V4-Pro Preview despite a much smaller activated parameter count. [^2] A monitored analysis attributes the jump to post-training rather than a larger model: architecture and parameter scale remain unchanged, while Terminal-Bench rises from 61.8 to 82.7 and DeepSWE from 7.3 to 54.4. [^3] Artificial Analysis scores it 50—10 points above the previous Flash and one behind GPT-5.6 Luna—with GDPval-AA v2 rising from 1189 to 1559 and Terminal-Bench reaching 79%. [^1] At $0.14/$0.28 per million input/output tokens, with a 1M-token context, Artificial Analysis estimates roughly 60% lower cost per task than Luna even after OpenAI’s 80% price cut. [^1]

**MiniMax H3 extends the open-model challenge into video.** A monitored launch summary reports H3 at #1 in video editing, #2 in text-to-video, and #3 in image-to-video, with 5–15-second native-2K, 24fps clips and stereo audio at $7.80 per minute versus $22.45 for Seedance 2.0 and $20.16 for Kling 3.0. [^4] MiniMax says the model is priced for production and will be open for anyone to build on within days. [^5]

## Research & Innovation

*Why it matters: The strongest new signals test provenance, scaffolding, and repeatability—not just headline scores.* [^6]

**ConjectureBench makes frontier-math claims easier to check.** Bespoke Labs lists a Jacobian-conjecture disproof attributed to Claude Fable 5, a Maxwell-conjecture disproof assisted by GPT-5.6 Sol, and an independently verified cycle-double-cover proof from GPT-5.6; its new repository collects 15,000 source-linked open problems for model-and-human investigation. [^7]

**ReviewBench shows that agent design can dominate model choice.** LangChain converted real reviewer comments from merged PRs into 59 Harbor tasks covering 64 issues. Basic harnesses recover only about 30% of curated findings, but a structured review prompt—without new tools—raised Luna to 0.32 on a 20-task slice, above static Kimi and Opus runs. [^6]

## Products & Launches

*Why it matters: Assistants are moving from chat windows into persistent workflows and the browser.* [^8][^9]

**Google expanded Gemini’s workflow layer.** Gemini 3.6 Flash and 3.5 Flash-Lite are available with claimed reasoning and speed improvements; Spark is rolling out to more countries and languages as a 24/7 personal agent, while Gemini gains Dropbox, Viator, and Zillow connections. [^10][^8][^11]

**ChatGPT is becoming more web-native.** Its Chrome extension can discuss YouTube videos, open tabs, and highlighted page text; the desktop app adds URL suggestions and browser-history controls. [^9]

## Industry Moves

*Why it matters: Deployment economics are translating into both massive infrastructure commitments and new access rules.* [^12]

**Amazon raised its 2026 capex guidance to $220 billion.** Andy Jassy’s reported case is that data centers are two-year builds with decades of revenue, while equipment pays back in about three years and lasts five to six; he said demand will exceed capacity in 2026 and 2027, with 2028 reservations already arriving. [^12]

**Efficiency is not replacing scale.** A Bloomberg-sourced report says DeepSeek is seeking 1 GW of compute in Ulanqab, alongside its aggressive model efficiency push. [^13] Meanwhile, Thinking Machines argues that neither indiscriminate weight release nor keeping capable models inside a few labs is safe, proposing staged access for Inkling. [^14]

## Quick Takes

*Why it matters: The edge is spreading across agent loops, professional reliability, speech, and video.*

- **OpenMLE** released a full-stack recursive-self-improvement testbed; its Frontis-MA1 agent raised MLE-Bench Lite Medal Average from 39.39% to 60.61%, reaching 71.21% with asynchronous search. [^15]
- **APEX-Accounting** found no tested model can reliably close the books: the leader scored 56.4%, 58% of tasks were never solved, and all-eight-run success reached only 2.6%. [^16]
- **Qwen-Audio-3.0-ASR-Flash** reports internal medical-term recall of 95.36% and industrial-term recall of 93.24%, with hotwords and structured transcript polishing. [^17]
- **Grok Imagine Video 1.5** added text-to-video, image and voice references, and native 1080p to its API and consumer products. [^18]

---

### Sources

[^1]: [𝕏 post by @ArtificialAnlys](https://x.com/ArtificialAnlys/status/2083123180869496865)
[^2]: [𝕏 post by @kimmonismus](https://x.com/kimmonismus/status/2083177904616202470)
[^3]: [𝕏 post by @ZhihuFrontier](https://x.com/ZhihuFrontier/status/2083191714874048833)
[^4]: [𝕏 post by @MTSlive](https://x.com/MTSlive/status/2083262338514456661)
[^5]: [𝕏 post by @MiniMax_AI](https://x.com/MiniMax_AI/status/2083373860796153869)
[^6]: [𝕏 article by @LangChain](https://x.com/i/article/2083185690137792512)
[^7]: [𝕏 article by @bespokelabsai](https://x.com/i/article/2083209815623712768)
[^8]: [𝕏 post by @GeminiApp](https://x.com/GeminiApp/status/2083233594395517183)
[^9]: [𝕏 post by @ChatGPT](https://x.com/ChatGPT/status/2082970812584432115)
[^10]: [𝕏 post by @GeminiApp](https://x.com/GeminiApp/status/2083233596362600840)
[^11]: [𝕏 post by @GeminiApp](https://x.com/GeminiApp/status/2083233600238166100)
[^12]: [𝕏 article by @jaminball](https://x.com/i/article/2083004147406983168)
[^13]: [𝕏 post by @kimmonismus](https://x.com/kimmonismus/status/2083192935852491096)
[^14]: [𝕏 post by @thinkymachines](https://x.com/thinkymachines/status/2083338736436400536)
[^15]: [𝕏 post by @dair_ai](https://x.com/dair_ai/status/2083231722913882159)
[^16]: [𝕏 post by @mercor_ai](https://x.com/mercor_ai/status/2083235089027399766)
[^17]: [𝕏 post by @Alibaba_Qwen](https://x.com/Alibaba_Qwen/status/2083111834123407825)
[^18]: [𝕏 post by @grok](https://x.com/grok/status/2083353607370416632)