🚨 DeepSeek V4 (GA) is reportedly in testing.
Early reports suggest:
• Opus 4.8-level performance
• Much stronger agentic behavior
• Significant coding improvements
• Major gains in 3D generation
• Direct competitor to Kimi K3 and Claude Fable
• Rumored pricing: $0.0028 pricing is accurate, DeepSeek V4 could be orders of magnitude cheaper than today's frontier models while delivering near-frontier intelligence.
This could be one of the most disruptive AI releases we've ever seen.
Performance is becoming a commodity.
Cost will become the battlefield.
Early reports suggest:
• Opus 4.8-level performance
• Much stronger agentic behavior
• Significant coding improvements
• Major gains in 3D generation
• Direct competitor to Kimi K3 and Claude Fable
• Rumored pricing: $0.0028 pricing is accurate, DeepSeek V4 could be orders of magnitude cheaper than today's frontier models while delivering near-frontier intelligence.
This could be one of the most disruptive AI releases we've ever seen.
Performance is becoming a commodity.
Cost will become the battlefield.
Jul 18, 2026 · 3:04 PM UTC
157
269
3,159
504,656
@EntelligenceAI AI competition is accelerating at lightning speed
1
1
6
981
@EntelligenceAI Will believe when it comes out or announced by DeepSeek
0
0
2
626
@EntelligenceAI Some free API credits to test: x.com/OrcaRouter/sta…
🚀@TencentHunyuan Hy3 is now FREE on OrcaRouter
295B MoE, 21B active parameters, 256K context, built for reasoning, coding, long-context work & agents.
Invite friends and claim the launch offer 👇
orcarouter.ai/redeem/HY3WITH…
#Hy3 #LLM
0
0
1
34
@EntelligenceAI “Opus level performance” “direct competitor to Fable” right..
0
0
1
397
@EntelligenceAI I run V4 Pro now and it's awesome (the version that is available for like 6 months now).
Best price for performance.
Best price for performance.
3
0
40
4,864
@EntelligenceAI 'reportedly' and 'rumored' doing a lot of heavy lifting here
ill believe the benchmarks when someone outside the hype cycle runs them
ill believe the benchmarks when someone outside the hype cycle runs them
0
0
37
3,174
@EntelligenceAI Sir, the Chinese have deployed AI Chips...and they aren't GPU's, TPU's or NPU's. How should we respond?
1
3
22
2,705
@EntelligenceAI rumored pricing that low would undercut nearly everyone if it holds up.
0
0
21
2,108
@EntelligenceAI Wait, WHAT, HOW THE FUCK IS THIS POSSIBLE.
1
0
14
5,474
@EntelligenceAI $0.0028 per 1M tokens is insane, the pricing war is actually eating itself
frontier in 6 months is gonna be like $0.00003 and a free t-shirt
frontier in 6 months is gonna be like $0.00003 and a free t-shirt
0
0
12
656
@EntelligenceAI Been using it as my daily driver, just via paid API, not really doing anything different but the speeds and quality boost in the last 48 hours have been phenomenal.
I do a lot of front end and I'd say it was indeed up there with Opus 4.8 all of a sudden.
Just my IRL experience.
I do a lot of front end and I'd say it was indeed up there with Opus 4.8 all of a sudden.
Just my IRL experience.
1
0
6
2,294
@EntelligenceAI What utter AI generated drivel ...
It's the same model FFS. Most they can do is fix tool calling in the same way Google did with Gemma 4, after lots of community feedback.
It's the same model FFS. Most they can do is fix tool calling in the same way Google did with Gemma 4, after lots of community feedback.
3
0
5
5,141
@EntelligenceAI If they "open source" this, it's going to be huge
0
0
3
635
@EntelligenceAI imma lose my shit if it is actually priced at $0.0028
0
0
3
403
@EntelligenceAI Opus Niveau, bessere Agenten, 3D Sprung und Spottpreis. Das ist kein Benchmark, das ist bereits die komplette Speisekarte. Spannend wird es erst, wenn offene Gewichte, reproduzierbare Tests und echte API Antworten zusammenpassen. Vorher ist V4 vor allem sehr gutes Marketing.
0
0
2
3,449
@EntelligenceAI Of that pricing is correct , we can expect to see red all over the place next week
1
0
3
988
@EntelligenceAI Rumored models are undefeated. Every leak beats the current champion in every category, and then the released version shows up needing a patch. Reportedly is doing all the benchmarking here.
0
0
2
979
@EntelligenceAI I use DeepSeek v4 Pro to create python tools, extensions. This ai provide very big codes
0
0
2
180
@EntelligenceAI the fuck you mean 2.8 cents per 10 million tokens 😭
0
0
2
69
@EntelligenceAI the pricing is confirmed by deepseek themselves its on their site, its not gonna change for GA
0
0
2
695
@EntelligenceAI Are you fucking kidding me? If these numbers are real it's actually over for most model companies
0
0
2
1,012
@EntelligenceAI Vision I’m calling it because they pulled their vision docs down
0
0
2
1,576