Pangram Raises $9M, Launches Next-Gen Text and Image AI Detectors
AI content detection startup Pangram announced a $9 million funding round led by Menlo Ventures, with participation from Haystack, ScOp, and others. On the same day, the company released two new detection models, Pangram 4 and Pangram Image, to address the growing need for identifying AI-generated content.
Confirmed
- Funding Details: Pangram closed a $9 million round led by Menlo Ventures, with Deedydas also participating.
- Pangram 4 Text Detection Model: This model's parameter count is 6 times larger than its predecessor. Architecturally, it incorporates a token-wise head into the classifier. The company states that the model exhibits enhanced robustness against "humanizer" rewriting tools, mixed human writing, and content generated by the latest frontier models.
- Pangram Image Detection Model: Launched as a public research preview to identify AI-generated images. The company claims a commercial benchmark accuracy of 99.5%, capable of detecting images generated by mainstream models including GPT Image, Gemini's Nano Banana, and Midjourney.
Why It Matters
- Background & Need: Relevant data indicates that AI-generated content now accounts for 35-50% of internet content, causing the cost of content trust to skyrocket. Pangram aims to resolve the trust crisis triggered by the proliferation of AI content by providing highly accurate detection tools.
2026-07-29 ~ 2026-07-30 · 6 related posts
Primary sources
- [source] Pangram raises $9M to detect AI text and image content as web slop grows — RebeccaBellan · 2026-07-29
- Pangram launches version 4 of its AI text detector — ctjlewis · 2026-07-29
- Pangram unveils Pangram 4 and Pangram Image for text and image AI detection — TuhinChakr · 2026-07-29
- [source] Pangram Raises New Round, Launches Stronger AI Text and Image Detection Models — deedydas · 2026-07-29
- Pangram launches image-detection preview claiming 99.5% accuracy on commercial benchmarks — nrehiew_ · 2026-07-30
- [source] Pangram 4 AI Detector Achieves 99.66% Accuracy with 0.0041% False Positive Rate — IgorBrigadir · 2026-07-30