Spark-X2.6-4B Released with Stronger Coding and Agent Skills
SparkLLM released Spark-X2.6-4B, a 4B model with improved post-training for coding, agent tasks, and long-context handling, offering three thinking levels.
Global AI Digest
-
-
TokenSpeed Adds Day-0 Support for MiniMax M3.1 Flash
LightSeek's TokenSpeed now supports MiniMax M3.1 Flash from day one, featuring DSpark speculative decoding and MXFP8 + NVFP4 mixed precision on Blackwell, and i…
-
MiniMax M3.1 Flash Open Weights Released, Live on Together AI From Day 0
MiniMax released open weights for M3.1 Flash, with Together AI as a day-0 launch partner offering fully managed production inference. The model targets agentic…
-
MiniMax Partners With Inferact on Production Inference for M3.1 Flash
MiniMax announced a partnership with Inferact to advance production inference for its models, with engineering improvements contributed back to the vLLM project…
-
MiniMax M3.1 Flash Hits 1K+ tok/s With Day-0 SGLang Support
MiniMax M3.1 Flash achieves over 1K tok/s at batch size 1 decode with SGLang, with day-0 production support across NVIDIA Blackwell/Hopper and AMD hardware, plu…
-
MiniMax-M3.1-Flash Gets Day-0 Fine-Tuning Support in NVIDIA NeMo
MiniMax-M3.1-Flash is now open weights and has day-0 fine-tuning support in NVIDIA NeMo-AutoModel, including FSDP2, expert, pipeline and context parallelism sca…
-
OpenRouter Model Aliases Auto-Track the Latest Versions
OpenRouter supports "-latest" model aliases such as ~anthropic/claude-opus-latest, so code automatically picks up the newest version of a model family without c…
-
MiniMax M3.1 Flash Releases Open Weights: 428B Parameters, 23B Active
MiniMax has published the weights of M3.1 Flash, a 428B-parameter model with 23B activated per token, reporting benchmark scores including 76.2 on BankerToolBen…
-
You.com Launches Inference Router with GLM-5.3 and Other Models
You.com introduced Inference Router, an enterprise-grade on-demand inference service offering Kimi K3, GLM-5.3, DeepSeek V4.1 Flash, Qwen3.8 Flash and other mod…
-
OpenRouter Security Center Adds Key Spend Alerts and Lifetime Limits
OpenRouter's gateway Security Center now lets users audit all API keys in one place, set spend alerts, remove unused keys, cap spending on uncapped keys, and en…
