Google and DeepSeek release new Flash and Pro models
Google/DeepMind
DeepSeek
OpenAI
Anthropic
Cerebras Systems
Moonshot AI
xAI
Google launched Gemini 3.7 Flash, claiming it's its most powerful Flash model for coding and agentic tasks, and halved prices for the current Flash generation. DeepSeek released V4-Pro in general availability with new peak/off-peak pricing, while OpenAI and Cerebras introduced an 'Ultrafast' tier for GPT-5.6 Sol.
Google unveiled Gemini 3.7 Flash just three weeks after Gemini 3.6 Flash, calling it its most capable Flash model for coding and agentic tasks. In Google's benchmarks, the model leads in FrontierCode 1.1 Main with 43.6%, but trails GPT-5.6 Terra in DeepSWE v1.1 (65.3% vs 69.6%). Google halved prices for the entire current Flash generation, with Gemini 3.7 Flash priced at $0.75 per million input tokens and $3.75 per million output tokens until December 31, 2026. DeepSeek released the general availability version of V4-Pro, introducing adjustable reasoning intensity and peak/off-peak pricing where off-peak is 50% cheaper, effective from August 16. Meanwhile, OpenAI and Cerebras introduced an 'Ultrafast' service tier for GPT-5.6 Sol, claiming up to 750 output tokens per second on Cerebras' Wafer Scale Engine hardware, a 14x improvement over standard mode, available initially in a limited preview.
- Abbreviations
- API = Application Programming Interface
- GA = General Availability
- GPU = Graphics Processing Unit
- SRAM = Static Random-Access Memory
- UTC = Coordinated Universal Time
Source: Heise online —
original