DeepSeek cut input prices on several models, with V4 Pro dropping 74% to $0.41 per 1M tokens. The broader ecosystem saw 149,514 GitHub stars gained across tracked projects today, with 68 new projects entering the index. This week added 2,120 new projects and 100 new Hugging Face spaces. Public-apis led daily star gain with +5,661.
AI News · August 23, 2026
DeepSeek slashes prices, Ray 2.58 adds KV cache routing
149,514 stars today68 new projects+5,661 public-apis
Releases of the day
Cline cli-v3.0.57 added hub drain, which stops a hub from accepting new mutating work while it finishes existing tasks, with a --off flag to lift the drain. LiteLLM v1.98.0 now signs all Docker images with cosign, and every release uses the same key introduced in commit 0112e53 for verification. Ray 2.58 completed KV cache and token aware request routing, previewed in 2.57, with tokenization now happening in-process. The Ray Serve release mirrors these highlights, improving LLM serving efficiency through better routing decisions.
Cline cli-v3.0.57 Added cline hub drain, which stops a hub from accepting new mutating work while it finishes what it is already running, and cline hub drain --off to lift it - Added cline hub...
LiteLLM v1.98.0 Verify Docker Image Signature · All LiteLLM Docker images are signed with cosign. Every release is signed with the same key introduced in commit 0112e53. · Verify using the...
Ray Serve ray-2.58.0 Highlights · * Ray Serve LLM: In this release we've completed KV cache and token aware request routing, which was previewed in 2.57. Tokenization now happens in-process on the...
Ray ray-2.58.0 Highlights · * Ray Serve LLM: In this release we've completed KV cache and token aware request routing, which was previewed in 2.57. Tokenization now happens in-process on the...
Pricing moves
DeepSeek V4 Flash 0423 input price fell 25% to $0.06 per 1M tokens, V4 Pro 0423 fell 74% to $0.41, and V3.2 fell 4% to $0.26. Kimi K2.6 from Moonshot AI dropped 7% to $0.54. On the upside, gpt-oss-120b from OpenAI rose 33% to $0.04, Qwen3 Next 80B A3B Instruct from Alibaba rose 11% to $0.1, DeepSeek V3.1 rose 120% to $0.55, and Muse Glimmer 30B from Meta rose 17% to $0.35.
▼ 25%DeepSeek V4 Flash 0423DeepSeek · input $0.08 → $0.06 per 1M tokens
▼ 74%DeepSeek V4 Pro 0423DeepSeek · input $1.6 → $0.41 per 1M tokens
▼ 4%DeepSeek V3.2DeepSeek · input $0.27 → $0.26 per 1M tokens
▼ 7%Kimi K2.6Moonshot AI · input $0.58 → $0.54 per 1M tokens
▲ 33%gpt-oss-120bOpenAI · input $0.03 → $0.04 per 1M tokens
▲ 11%Qwen3 Next 80B A3B InstructAlibaba · input $0.09 → $0.1 per 1M tokens
▲ 120%DeepSeek V3.1DeepSeek · input $0.25 → $0.55 per 1M tokens
▲ 17%Muse Glimmer 30BMeta · input $0.3 → $0.35 per 1M tokens
Emerging projects
Guizang-ppt-skill entered the index today with 24,632 stars. It is an HTML-based AI-agent skill for generating polished HTML slide decks, supporting editorial magazine and Swiss layouts, image prompts, and social covers.
The bigger picture
Price cuts on DeepSeek models contrast with increases on several others, signaling competitive pressure in inference pricing. Ray's KV cache routing advances LLM serving, while LiteLLM's image signing improves security. The ecosystem continues to expand rapidly with many new projects entering the index.
Source: olud.ai tracking of 10,000+ open-source AI projects, 300+ models and live provider pricing. Figures are measured, not estimated. All releases · Live pricing · Latest in AI
← All AI news