Observed Signal · Jul 15, 2026 · Product Launch · Source: techcrunch · Impact: 3/5 · Sentiment: Positive
Large Language Models (LLM) & AI Market: Thinking Machines releases open-weight model Inkling
Thinking Machines released Inkling, an open-weights, general-purpose multimodal Mixture-of-Experts foundation model: a 66-layer decoder-only transformer with a sparse MoE backbone (≈975B total parameters, ≈41B active per task). It natively accepts text, images and audio and produces UTF-8 text outputs. Trained from scratch on roughly 45 trillion tokens assembled from public, third-party and synthetic sources on NVIDIA GB300 NVL72 systems under a strategic NVIDIA partnership, Inkling ships with checkpointed support for up to a 1,000,000-token context (Tinker API exposes 256K) and includes a smaller preview (Inkling‑Small, ≈276B/12B). Published under Apache‑2.0 with day‑0 ecosystem and inference/hosting support (Hugging Face, vLLM, SGLang, TokenSpeed, Modal, Databricks, Baseten, NVIDIA optimizations and community quantization), it is positioned for enterprise fine‑tuning via Tinker, evaluated against benchmarks (2026-07-14) and accompanied by safety guidance.
An open-weight, multi-modal 975B-parameter model from a well-funded startup with an Nvidia partnership could accelerate enterprise customization and open-source model adoption, but it is not a major platform (Google/Meta/Apple) policy shift.
Key Takeaways & Evidence Grounding
- Architecture: 66-layer decoder-only transformer with a sparse MoE backbone — ≈975B total parameters, ≈41B active per task; accepts text, image and audio inputs and outputs UTF-8 text.
- Training: trained from scratch on roughly 45 trillion tokens drawn from public, third-party and synthetic/augmented datasets; training ran on NVIDIA GB300 NVL72 systems under a strategic NVIDIA partnership.
- Context & releases: checkpointed support for up to a 1,000,000-token context window; Thinking Machines’ Tinker API currently exposes 256K; preview Inkling‑Small (~276B/12B) available.
- License & ecosystem: published under Apache‑2.0 with day‑0 integration and local deployment support (Hugging Face, vLLM, SGLang, TokenSpeed, Modal, Databricks, Baseten), plus NVIDIA optimizations and community quantization; API access via third-party providers.
- Positioning & evaluation: positioned as an enterprise-customizable base model fine-tuned via Tinker; benchmark comparisons reported (2026-07-14) and releases include safety mitigation recommendations.
Connected Companies & Entities
9 Entities mappedAnthropic
Foundation model company selling AI assistants and model APIs.
“Thinking Machines Lab, the AI startup founded by former OpenAI CTO Mira Murati, released its first proprietary AI model Wednesday morning, c...”
NVIDIA
Accelerated computing company spanning AI software, cloud and gaming.
“It struck a strategic partnership with Nvidia in March to deploy a gigawatt of Vera Rubin computing capacity, and says Inkling itself was tr...”
Thinking Machines Lab
Builds multimodal foundation models and fine-tuning infrastructure.
“Thinking Machines Lab, the AI startup founded by former OpenAI CTO Mira Murati, released its first proprietary AI model Wednesday morning, c...”
OpenAI
Foundation model company selling AI software, APIs and subscriptions.
“Thinking Machines Lab, the AI startup founded by former OpenAI CTO Mira Murati, released its first proprietary AI model Wednesday morning, c...”
Search, video, adtech and cloud giant within Alphabet.
“Thinking Machines Lab, the AI startup founded by former OpenAI CTO Mira Murati, released its first proprietary AI model Wednesday morning, c...”
Hugging Face
Open AI model hub with hosted inference and collaboration.
“Hugging Face CEO Clem Delangue made a similar prediction in conversation with TechCrunch last week....”
Microsoft
Diversified software, cloud, advertising and gaming platform company.
“In a blog post published Sunday, Microsoft CEO Satya Nadella — whose company has invested billions in both OpenAI and Anthropic — warned tha...”
Moonshot AI
Foundation model company behind Kimi consumer and enterprise AI.
“Thinking Machines pretrained Inkling from scratch, but it says it used other open-weight models — including Moonshot AI’s Kimi K2.5 — to hel...”
Moonshot AI
Chinese AI company delivering Kimi models, agents, enterprise software and APIs.
“Thinking Machines pretrained Inkling from scratch, but it says it used other open-weight models — including Moonshot AI’s Kimi K2.5 — to hel...”
Ontology Mapping & Concepts
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.