AI Daily · Oct 28, 2026: GPT-6.1 Terra lands on B.AI with 1.05M context at $1/$5 per million tokens
OpenAI's GPT-6.1 Terra is live on B.AI with a 1.05M context window and adjustable reasoning at $1/$5 per million tokens, while Anthropic offers Claude to US governments at 40% off under OneGov.
Pick a date
-
GPT-6.1 Terra launches on B.AI with 1.05M context and $1/$5 per million tokens
OpenAI's GPT-6.1 Terra is live on B.AI's API and web chat, with adjustable reasoning effort, a 1.05M context window, up to 128K output tokens, and standard pricing of $1.00 per million input and $5.00 per million output tokens.
-
Anthropic offers Claude to US governments at 40% discount under OneGov
Anthropic expanded its partnership with the US GSA via a new OneGov agreement, giving federal, state, local, tribal and territorial governments 40% off Claude.
-
Liquid AI releases LFM3 family, now running locally via Ollama
Liquid AI introduced LFM3, a new generation of Liquid Foundation Models with a tri-hybrid architecture and multimodal encoders, spanning 6 model instances across text, audio, vision and omni modalities, including LFM3-1.7B for text; the family is now on Ollama, where the text model fits in about 1 GB at 4-bit with a 256K context window.
-
ByteDance Seedance video generation series launches on B.AI API
ByteDance's Seedance video models are now available via the B.AI API: Seedance 2.5 generates clips up to 30 seconds at 1080p 10-bit with up to 50 reference inputs and timestamp editing; Seedance 2.0 supports up to 4K 10-bit output; Seedance 2.0 Fast targets 480p/720p at lower cost.
-
Command Code offers $80 of DeepSeek or GLM usage for $10 monthly
Command Code's GOAT plan costs $10 per month and includes $80 of usage on DeepSeek or GLM models; the post also claims DeepSeek v4.1 flash 1017 is the best open-weights model currently available.
-
AssemblyAI launches its first text-to-speech model with 90 ms streaming
AssemblyAI-TTS-1 is the company's first TTS model, aimed at voice agents, streaming at 90 ms latency and priced at $15 per million characters.
-
MiniMax M3.1 Flash free on EvoX Agent from Oct 28 to 30
MiniMax M3.1 Flash is offered free on EvoX Agent for Oct 28–30; the post notes an earlier version of M3.1 Flash topped global leaderboards with 200T+ tokens.
-
QoderBench: GLM-5.3 at High cuts cost 40.5% versus Max with 6.3% score drop
On QoderBench's nearly 200 cases, running GLM-5.3 at High instead of Max reduced cost per run by 40.5% while score fell 6.3% relative to Max.
-
Mistral serves DeepSeek V4.1 Flash and GLM 5.3 Flash
Mistral now serves two low-cost models: DeepSeek V4.1 Flash in public preview for long-context coding and agentic tasks, and GLM 5.3 Flash now generally available for subagents, tool calls and coding.
-
OpenAI's Codex/ChatGPT day 23 brings a usage reset
OpenAI's Tibo announced a usage reset for Codex and ChatGPT on day 23, noting iCloud Mail and Calendar plugins remain a requested feature.