Open source AI
Latest stories covering Open source AI, with summaries and links to related coverage.
2 stories · 2 reports · page 1 of 1
1 story · 1 report in the last 7 days
Latest report · Counts as of (UTC)
Models in this coverage
Up to six tracked models linked to this topic’s reports in the last 30 days. Counts measure coverage on Themata, not usage.
- DeepSeek V4.1 Flash
DeepSeek
· Open weights
An open-weight mixture-of-experts model that reads text and images. Its architecture focuses on efficient long-context and agent workloads, with support for up to one million tokens.
1 story · 1 report on this topic
No established topics share multiple stories with this topic yet.
Latest stories
The same AI request emits 8.5 times more CO2 in Texas than in Sweden
tool
6 min read
Berget AI released an open-source method for calculating the carbon emissions of individual AI inference requests, developed with the Stockholm Environment Institute, WattTime and Klimatkollen. The Swedish AI infrastructure provider has added per-request emissions tracking to its customer console and published its calculator and methodology under an MIT licence. The method follows the Green Software Foundation’s SCI for AI specification and accounts for infrastructure energy use, cooling and hardware manufacturing, while excluding model training. Using identical hardware, Berget AI estimates that a typical Kimi K3 request emits 0.38 grams of CO₂e in Sweden, 1.6 grams in Ireland, 2.5 grams in Germany and 3.2 grams in Texas. The Texas estimate is 8.5 times Sweden’s, chiefly because of differences in grid emissions intensity. The example request uses about 4 Wh of energy, consists of 5,000 input and 387 output tokens, and is measured on a shared B300 deployment. Hardware manufacturing represents 89% of the Swedish result, while electricity represents 90% of the Texas result; the Texas request also uses about 6 mL of cooling water. The calculator applies location-based average grid emissions and can estimate any model and region. Berget AI said it will add marginal-emissions estimates next month and expand coverage to chip-manufacturing water and energy impacts. The company corrected earlier Ireland, Germany and Texas estimates on 21 October 2026 after reviewing its grid-intensity sources.
DeepSeek just open-sourced its inference engine
tool
2 min read
DeepSeek has open-sourced DeepInfer, the high-performance inference framework used in its large-scale production services. The release supports offline inference for DeepSeek-V4.1-Flash and includes optimized kernels and model and sampler runtimes. Its parallelism features include attention data parallelism, balanced context parallelism, mixture-of-experts expert parallelism, and expert parallel load balancing. DeepInfer uses FlashMLA, DeepGEMM, DeepEP, TileKernels, DeepSelect, and custom TileLang kernels. The framework supports DSpark and confidence-scheduled verification to accelerate decoding with draft tokens, decoder sliding-window-attention bounded replay to reduce prefill work, and asynchronous scheduling that overlaps CPU batch scheduling with device execution. It can provide batch-invariant inference and deterministic sampling on NVIDIA hardware, and it can constrain outputs with regular expressions, JSON schemas, or EBNF grammars. Supported hardware includes NVIDIA SM100 GPUs and Huawei Ascend 950 NPUs. DeepInfer requires Linux, GCC 13 or newer, and Python 3.12, plus CUDA 13 for NVIDIA or CANN 9.2 for Ascend. A complete serving system still requires components such as a KV-cache system, prefill/decode disaggregation, and load balancing across data-parallel replicas. DeepInfer is available under the Apache License 2.0.