DeepSeek announced new peak/off-peak pricing for its DeepSeek-V4-Flash and DeepSeek-V4-Pro models, effective August 17, 2026, with off-peak input (cache miss) prices at 1.5 yuan and 4.5 yuan per milli…
DeepSeek is raising prices for its flagship V4 models by more than four times, with new peak-hour pricing taking effect Aug. 16. The Hangzhou-based AI lab will charge $1.32 per 1 million output tokens…
A developer released squint-mcp, an open-source Model Context Protocol server that converts images into ASCII art so text-only large language models like DeepSeek-V4-Flash and GLM-5.2 can attempt to i…
Nicholai Mitchko, InterSystems' director of AI enablement, released DeepSeek-V4-Flash-0731-Latent-Reasoning, a self-hosted latent-reasoning model built on DeepSeek-V4-Flash that requires Blackwell-cla…
A new benchmark called ConstraintRot, detailed in the paper *Governance Decay* (arXiv:2606.22528), found that context compaction caused an agent to violate a prohibited action 59% of the time on DeepS…
A developer cut a nightly batch classification job from 40 minutes to 6 minutes by switching to DeepSeek-V4-Flash and adopting an async request pattern. The model tier swap and concurrency changes wer…
China's National Supercomputing Internet launched API access and model downloads for DeepSeek-V4-Flash on August 4, 2026, marking the model's public beta debut. The integration provides developers wit…
DeepSeek-V4-Flash entered public beta, with China's National Supercomputing Internet offering API access and model downloads on day one. Developers can use the API through the platform's Model Service…
DeepSeek-V4-Flash, the efficiency tier of DeepSeek's V4 series, costs $0.28 per million output tokens, compared to about $25 for Claude Opus 4.8, and performs within a few points of it on agentic codi…
DeepSeek released DeepSeek-V4-Flash in public beta on 2026-07-31, showing significantly enhanced agent capabilities with benchmark results far exceeding V4-Pro-Preview, including Terminal Bench 2.1 at…
An experiment by an unnamed researcher found that retrying failed requests in LLM benchmarks introduces selection bias, shortening average response lengths by 19.2% and reducing long-form outputs by u…
Doubleword, one of six companies in the first wave of UK Sovereign AI investments, achieved up to 3× the throughput of vLLM for DeepSeek-V4-Flash on a single node of Isambard-AI, the UK's national AI …
A developer's analysis of coding agent costs reveals a 63× price spread across models, from $19/month for Qwen3.5-Flash to $1,200/month for GPT-5.6 Sol, based on a fixed workload of 90M input and 25M …
Kimi-K2.7-Code defeated DeepSeek-V4-Flash in a head-to-head benchmark with an aggregate score of 112.4 to 99.8, winning 5 tasks to 3 with 4 ties and a 75% confidence lean. The evaluation, conducted by…
A developer successfully ran DeepSeek-V4-Flash, a 284B-parameter mixture-of-experts model, on a 64GB AMD Strix Halo laptop by streaming cold experts from SSD via mmap. The model achieves ~1.9 tok/s de…
AutoTrainess, a language model training agent, outperforms traditional CLI methods by automating training workflows, achieving a 26.94 average score on PostTrainBench with GPT-5.4 (Codex) versus 23.21…
DeepSeek AI released preview versions of its DeepSeek-V4 series, including two Mixture-of-Experts language models with up to 1.6 trillion parameters and support for one-million-token contexts. The mod…
A developer cut their AI API bill by 40% without changing application code by switching to a gateway that normalizes multiple providers to the OpenAI API format. By changing only the base_url and api_…
Princeton University's Language and Intelligence Lab published a paper introducing Goedel-Architect, an agent framework for formal theorem proving built around DeepSeek's open-source V4-Flash model. O…
A university login server at IIT Delhi that appeared secure after blocking a public Dirty Frag exploit was compromised in approximately 90 minutes using a DeepSeek-V4-Flash automated feedback loop. Th…