AI Intelligence Digest
TIMPS
PostCards
The Open-Source Surge: Why Big Labs Are Losing Their Moat
Issue 046 May 9, 2026 Hyderabad, India
⬡ Built with the TIMPS Ecosystem
Latest Issue
01 · LEAD
Open Source

DeepSeek V3.2 and GLM-5.1 Match Frontier Proprietary Models at a Fraction of the Cost

Open-source AI models have crossed a threshold: DeepSeek V3.2 delivers ~90% of GPT-5.4's performance at 1/50th the price, while GLM-5.1 (MIT license) tops SWE-bench Pro — and both run on hardware available to any startup with a GPU budget.
1/50th DeepSeek V3.2 cost vs GPT-5.4

The open-source AI ecosystem reached an inflection point this week as multiple independent evaluations confirmed that DeepSeek V3.2, GLM-5.1 (MIT license), and Gemma 4 (Apache 2.0) collectively match or exceed proprietary frontier models in key benchmarks. DeepSeek V3.2 delivers approximately 90% of GPT-5.4's performance at roughly 1/50th the inference cost. GLM-5.1, a 744B MoE model (40B active) released under MIT license, topped SWE-bench Pro — the coding benchmark that measures real-world software engineering ability — surpassing both GPT-5.4 and Claude Opus 4.6.

The implications for AI business models are profound. LLM Stats logged 255 model releases from major organizations in Q1 2026 alone. Any company with GPU access can now run frontier-adjacent models without a single API call to the major labs. Google's Gemma 4 family — released under Apache 2.0, the first Gemma under an OSI-approved open-source license — has passed 400 million total downloads. The 31B Dense model runs on a single high-end GPU while competing with much larger proprietary models on reasoning benchmarks.

WhatLLM.org, BuildFastWithAI, LLM Stats, Google AI
02 · CAPITAL
Market Shift

Together AI Doubles Valuation to $8.3 Billion as Open-Weight Inference Cloud Booms

The open-weight inference cloud provider raised $800 million led by Aramco Ventures, with customer bookings crossing $1.15 billion in a single quarter as enterprises flee closed-model vendor lock-in.
$8.3B Together AI valuation; $800M round

Together AI announced an $800 million funding round led by Aramco Ventures, doubling its valuation to $8.3 billion — making it the highest-valued 'neocloud' provider exclusively focused on open-weight models. The company's customer bookings crossed $1.15 billion in Q1 2026 alone, driven by enterprises seeking alternatives to OpenAI and Anthropic's pricing. Together AI hosts over 200 open-source models including DeepSeek V3.2, Gemma 4, and Mistral Small 4, offering inference at rates that undercut closed APIs by 6-60x.

The valuation surge reflects a structural shift in the AI market: as open-weight models close the quality gap, enterprises are re-evaluating the economics of proprietary API dependency. Together AI's revenue is concentrated among large enterprises running production workloads on self-hosted or hybrid deployments. CEO Vipul Ved Prakash told TechCrunch that the company's growth 'confirms that open models are not just a research curiosity — they are the production infrastructure of the future.'

TechCrunch, Axios, Together AI
03 · ECOSYSTEM
Ecosystem

Mistral Small 4 Ships Under Apache 2.0 — 6.5B Active Parameters, Frontier-Adjacent Performance

Mistral's latest small model merges reasoning, vision, and coding into a single endpoint under Apache 2.0, available for free if self-hosted — and benchmarks show it outperforms several models 10x its size.
6.5B Active parameters; Apache 2.0

Mistral released Small 4 under Apache 2.0, a 6.5 billion active parameter model (from a 42B MoE base) that merges reasoning, vision, and coding into a single endpoint. Despite its small size, the model outperforms several models 10-20x its size on key benchmarks, including GPT-5.1 and Claude Sonnet 4.5 on coding and reasoning tasks. The model runs on consumer-grade GPUs and is free to self-host under the permissive Apache 2.0 license.

The release completes Mistral's strategy of offering capable models across every size tier, all under open licenses. Mistral Small 4 joins Mistral Large 3 and Mistral Medium as part of a family that collectively powers an estimated 15% of European AI startups. CEO Arthur Mensch positioned the release as 'proof that efficient architecture can overcome brute-force scaling' — a direct jab at the massive compute requirements of frontier models from OpenAI and Anthropic.

Mistral AI, BuildFastWithAI, WhatLLM.org
04 · SIGNALS
5 Things Worth Knowing

5 Key Signals

01
Alibaba's Qwen 3.5 (397B MoE, 17B active, 201 languages, Apache 2.0) positions China as a credible open-weight challenger
Alibaba Cloud ↗
02
Meta's Llama 4 open-weight family passes 500M downloads since release
Meta AI ↗
03
NVIDIA Nemotron 3 Super released as open-weight — first time NVIDIA releases a frontier-class model openly
NVIDIA ↗
04
Hugging Face reports 2.5M+ open models on platform, with 70% of new uploads being fine-tunes of open-weight bases
Hugging Face ↗
05
The 'build vs buy' calculus shifts: Gartner reports 34% of enterprises now run self-hosted open-weight models in production
Gartner ↗
05 · THEMES
This Week in Focus

Top Themes

Open-Source Parity
Inference Cost Arbitrage
Neocloud Economics
Model Efficiency Revolution
License Strategy as Moat
Tool of the Week
Together AI
Open-weight inference cloud hosting 200+ models at 6-60x below closed API pricing — $1.15B quarterly bookings.