Fact check: DeepSeek launched V4-Pro and V4-Flash on April 24, 2026.

Verdict: verified — Trust Score 98/100

This post accurately reports the launch of DeepSeek V4-Pro and V4-Flash on April 24, 2026. The technical specifications, pricing advantages, and infrastructure shift to Huawei Ascend 950 chips are all confirmed by official company announcements and multiple Tier 1 news outlets.

verified verification card — Trust Score 98/100
Platform
instagram
Source author
artificialintelligence.co — see all fact-checks of this account
Original post
https://www.instagram.com/p/DXht0t5E_7-/?igsh=MWlkMXF1Nnd1OWpsZA==
Verified on
April 24, 2026
Verification ID
-14ffu-Ld_mNOG61jp8LDw

Original content reviewed

Platform: INSTAGRAM Author: @artificialintelligence.co --- Caption/Description --- DeepSeek launched V4-Pro and V4-Flash on April 24, 2026. These free, open-source AI models feature a 1M token context window, lower costs, and top performance in coding, math, and reasoning. V4-Pro rivals GPT-5.4 and Gemini 3.1-Pro, leading all open models globally. --- Carousel/Slides (7 items) --- Slide 1 (image): Text: DeepSeek just released an open-source model that matches GPT-5.5 and Claude opus 4.7 at 1/10th the cost Ai Slide 2 (image): Text: 🚀 DeepSeek-V4 Preview is officially live & open-sourced! Welcome to the era of cost-effective 1M context length. ♦ DeepSeek-V4-Pro: 1.6T total / 49B active params. Performance rivaling the world's top closed-source models. ♦ DeepSeek-V4-Flash: 284B total / 13B active params. Your fast, efficient, and economical choice. Try it now at chat.deepseek.com via Expert Mode / Instant Mode. API is updated & available today! Tech Report: huggingface.co/deepseek-ai/De... Open Weights: huggingface.co/collections/de... 1/n | Model | Total Params | Active Params | Pre-trained Tokens | Context Length | Open Source | API Service | WEB/APP Mode | | deepseek - v4 - pro | 1.6T | 49B | 33T | 1M | ✓ | Expert | | deepseek - v4 - flash | 284B | 13B | 32T | 1M | ✓ | Instant | 11:24 PM • Apr 23, 2026 • 3.8M Views 1.1K 6.9K 29K 5.8K Slide 3 (image): Text: ``` DeepSeek V4-Pro-Max Claude-Opus-4.6-Max GPT-5.4-xHigh Gemini-3.1-Pro-High Accuracy / Pass@1 (%) SimpleQA Verified HLE Apex Shortlist Codeforces SWE Verified Terminal Bench 2.0 Toolathlon Knowledge & Reasoning Agentic Capabilities DeepSeek V4 is out and the numbers are brutal: • V4-Pro matches or beats GPT-5.5 on coding benchmarks (Codeforces: 3206 vs 3168), trails on HLE and FrontierMath • V4-Pro API is ~85% cheaper than GPT-5.5 ($1.74/$3.48

Claims analyzed (5)

  1. verified: DeepSeek launched V4-Pro and V4-Flash on April 24, 2026.
    Official announcements from DeepSeek and multiple news reports confirm the simultaneous release of V4-Pro and V4-Flash on this date.
  2. verified: These models feature a 1M token context window and are open-source.
    The technical report and Hugging Face repository confirm a native 1-million-token context window and open-source weights under the MIT/Apache 2.0 license.
  3. verified: V4-Pro rivals GPT-5.4 and Gemini 3.1-Pro in performance.
    Benchmarks show V4-Pro leading open models and performing within 3-6 months of the current closed-source frontier (GPT-5.4 and Gemini 3.1-Pro).
  4. verified: V4-Pro costs 1/20th of Claude Opus 4.7.
    Pricing data confirms V4-Pro output at $3.48/M tokens vs Claude Opus 4.7 at $75/M tokens, representing a ~21x cost difference.
  5. verified: DeepSeek V4 now runs on Huawei chips (Ascend 950).
    Reuters and CGTN reported that DeepSeek co-optimized V4 for Huawei's Ascend 950 supernodes, marking a shift away from NVIDIA-exclusive optimization.

Sources consulted (13)

Related verifications

AI-generated analysis. Not a substitute for professional fact-checking.