# GPT-5.4 - Out Now! **URL:** \ **Category:** Release Discussions\ **Tags:** openai\ **Created:** [March 5, 2026, 6:20pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744 "2026-03-05T18:20:47Z")\ **Posts on this page:** 20\ **Page:** 2
**Author:** ![valentinoPereira](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/valentinopereira/32/78806_2.png) [@valentinoPereira](https://forum.cursor.com/u/valentinoPereira)\ **Post date:** [March 6, 2026, 2:00pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/27 "2026-03-06T14:00:26Z")
Its too expensive to use for us legacy request based pricing users.. For a medium reasoning non-1million model - just 51k tokens consumes 2 requests - that’s unacceptable and hugely demotivating cursor team. ![Screenshot 2026-03-06 at 7.22.13 PM](https://us1.discourse-cdn.com/cursor1/original/3X/5/b/5b5236e1a94e776151fcd723f14ea3e6acf4c01c.png) ---
**Author:** ![Pat1](https://avatars.discourse-cdn.com/v4/letter/p/43a26b/32.png) [@Pat1](https://forum.cursor.com/u/Pat1)\ **Post date:** [March 5, 2026, 8:32pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/28 "2026-03-05T20:32:46Z")
I tried gpt 5.4 using legacy pro plan and got gutted with 60 request on one prompt that was fixing 2 issues. ![gpt54](https://us1.discourse-cdn.com/cursor1/original/3X/6/d/6dd01abbf2a2da3beacf876f01db04cd74505637.png) ---
**Author:** ![Zr93](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/zr93/32/10997_2.png) [@Zr93](https://forum.cursor.com/u/Zr93)\ **Post date:** [March 5, 2026, 8:53pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/30 "2026-03-05T20:53:24Z")
I think this is because 5.4 only has the max-mode, and I don’t know why non-maximum mode is unavaliable. ---
**Author:** ![valentinoPereira](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/valentinopereira/32/78806_2.png) [@valentinoPereira](https://forum.cursor.com/u/valentinoPereira)\ **Post date:** [March 6, 2026, 2:11pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/31 "2026-03-06T14:11:42Z")
yeah this is a big problem - stay away from this model - use Opus 4.6 thinking instead in non-max mode - it is capped at 2 requests ---
**Author:** ![glomar\_dev](https://avatars.discourse-cdn.com/v4/letter/g/f9ae1b/32.png) [@glomar\_dev](https://forum.cursor.com/u/glomar_dev)\ **Post date:** [March 6, 2026, 2:34pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/32 "2026-03-06T14:34:22Z")
![image](https://us1.discourse-cdn.com/cursor1/original/3X/d/b/dbc5b7c33b1db79763ae09631b44a0e0959a16f3.png) Seems like they’re trying to force legacy request-based users to move to token-based usage to me. ---
**Author:** ![Jackyfan](https://avatars.discourse-cdn.com/v4/letter/j/87869e/32.png) [@Jackyfan](https://forum.cursor.com/u/Jackyfan)\ **Post date:** [March 6, 2026, 3:03pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/33 "2026-03-06T15:03:59Z")
I’ve also run into the same issue. Sometimes subagents are triggered, and the sub-agent automatically calls the Composer 1.5 model, consuming 2 requests each time. ---
**Author:** ![AndreasM1](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/andreasm1/32/50651_2.png) [@AndreasM1](https://forum.cursor.com/u/AndreasM1)\ **Post date:** [March 6, 2026, 4:01pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/34 "2026-03-06T16:01:44Z")
Yes, this mapping of tokens to requests is totally exaggerated and simply unrealistic. In fact, they want the old legacy users to change their contracts or disappear completely. ---
**Author:** ![ryno](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/ryno/32/48905_2.png) [@ryno](https://forum.cursor.com/u/ryno)\ **Post date:** [March 6, 2026, 4:32pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/36 "2026-03-06T16:32:43Z")
Is GPT-5.4 unavailable within the CLI? It doesn’t seem to be listed as an available model. ---
**Author:** ![Scott1](https://avatars.discourse-cdn.com/v4/letter/s/7feea3/32.png) [@Scott1](https://forum.cursor.com/u/Scott1)\ **Post date:** [March 6, 2026, 5:36pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/37 "2026-03-06T17:36:29Z")
> [@G4Q4](#): > > Damn, that’s a bummer. I’m going to augment as they have 5.4 for free within their IDE If it will be like this, its time to change to Claude Code without this ■■■■■■■■■ We have the legacy pricing mode, because we ar long time customers. And we are being penalized for it? I’ll wait some days more to see if Cursor will stepback this nonsense. If not, bye bye. ---
**Author:** ![G4Q4](https://avatars.discourse-cdn.com/v4/letter/g/3e96dc/32.png) [@G4Q4](https://forum.cursor.com/u/G4Q4)\ **Post date:** [March 6, 2026, 6:39pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/39 "2026-03-06T18:39:52Z")
Why don’t you just use Opus 4.6 Thinking Max? That’s also only two requests. ---
**Author:** ![tmiley](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/tmiley/32/100732_2.png) [@tmiley](https://forum.cursor.com/u/tmiley)\ **Post date:** [March 6, 2026, 7:26pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/40 "2026-03-06T19:26:53Z")
Clearly this is gutting Legacy users. Why dont you just remove legacy mode instead of pretending you still support it? Stab us in the front instead of the back. ---
**Author:** ![sk12345](https://avatars.discourse-cdn.com/v4/letter/s/3ab097/32.png) [@sk12345](https://forum.cursor.com/u/sk12345)\ **Post date:** [March 6, 2026, 8:12pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/42 "2026-03-06T20:12:32Z")
Is the new computer use integrated? ---
**Author:** ![Colin](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/colin/32/90452_2.png) [@Colin](https://forum.cursor.com/u/Colin)\ **Post date:** [March 6, 2026, 10:38pm UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/43 "2026-03-06T22:38:17Z")
> [@ryno](#): > > Is GPT-5.4 unavailable within the CLI? It doesn’t seem to be listed as an available model. We’re adding it! Check back in ~30 minutes. ---
**Author:** ![NightStars](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/nightstars/32/2666_2.png) [@NightStars](https://forum.cursor.com/u/NightStars)\ **Post date:** [March 7, 2026, 4:01am UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/44 "2026-03-07T04:01:28Z")
We already lost valuable features (custom modes, memories), now we soon losing the value of legacy pricing..? I hope not, cuz that will straight up kill the product, the meaning and potential of Cursor AI, and that would be sad, I hope GPT 5.4 270K returns as normal and not max mode only, otherwise, a lot of us will reconsider ---
**Author:** ![NightStars](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/nightstars/32/2666_2.png) [@NightStars](https://forum.cursor.com/u/NightStars)\ **Post date:** [March 7, 2026, 4:02am UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/45 "2026-03-07T04:02:00Z")
Will GPT-5.4 be available as normal mode soon? ---
**Author:** ![M\_Yassine](https://avatars.discourse-cdn.com/v4/letter/m/7ab992/32.png) [@M\_Yassine](https://forum.cursor.com/u/M_Yassine)\ **Post date:** [March 7, 2026, 7:00am UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/46 "2026-03-07T07:00:38Z")
Honestly, even if I wanted to upgrade my plan after being forced to move from my current subscription, what guarantee do I have that this won’t happen again in the future? Cursor could simply change its subscription plans again, remove features I originally subscribed for, or make similar modifications—like restricting models to Max mode. At that point, it makes more sense for me to go directly to the source rather than rely on your product long-term. Since I’m now blocked from using the best model (GPT-5.4), I could simply subscribe to ChatGPT for $20/month and use it with Codex for coding, then use Opus 4.6 in Antigravity for my UI changes and it’s for free at that. ---
**Author:** ![Jackyfan](https://avatars.discourse-cdn.com/v4/letter/j/87869e/32.png) [@Jackyfan](https://forum.cursor.com/u/Jackyfan)\ **Post date:** [March 7, 2026, 10:18am UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/47 "2026-03-07T10:18:06Z")
Yes, we were among the earliest to support Cursor. I’ve already had a second-year annual subscription to Cursor. If Cursor changes the plan for old subscribers, it would be really disappointing for loyal users. Now, with some new features, they are indirectly consuming the credits of old users. This is very frustrating. ---
**Author:** ![JimKarter](https://avatars.discourse-cdn.com/v4/letter/j/b3f665/32.png) [@JimKarter](https://forum.cursor.com/u/JimKarter)\ **Post date:** [March 8, 2026, 8:13am UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/50 "2026-03-08T08:13:06Z")
Can you fix bad pricing instead of removing my comments, Colin? ---
**Author:** ![dylan\_wang](https://avatars.discourse-cdn.com/v4/letter/d/4491bb/32.png) [@dylan\_wang](https://forum.cursor.com/u/dylan_wang)\ **Post date:** [March 8, 2026, 10:25am UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/53 "2026-03-08T10:25:09Z")
Even though I specified in the configuration file that the subagent should call the Opus model, the cursor still calls the Composer model. Note that while Composer’s model capabilities are inferior, its price is twice that of Opus, and you cannot modify it. I posted this, and it was deleted. You never know what they’ll do next. For an unpredictable product user experience, I can only wish you good luck. Please try Codex and Claude Code. You’ll thank me. ---
**Author:** ![dylan\_wang](https://avatars.discourse-cdn.com/v4/letter/d/4491bb/32.png) [@dylan\_wang](https://forum.cursor.com/u/dylan_wang)\ **Post date:** [March 8, 2026, 10:33am UTC](https://forum.cursor.com/t/gpt-5-4-out-now/153744/55 "2026-03-08T10:33:19Z")
> [@Jackyfan](#): > > I’ve also run into the same issue. Sometimes subagents are triggered, and the sub-agent automatically calls the Composer 1.5 model, consuming 2 requests each time. Even if you specify it in the configuration file, it won’t work, and it’s twice as expensive as Opus. Welcome to Cursor. [Previous page](https://forum.cursor.com/t/gpt-5-4-out-now/153744.md?page=1) [Next page](https://forum.cursor.com/t/gpt-5-4-out-now/153744.md?page=3)