DEV Community: Aditya Vishwakarma

I Replaced $465/mo of Business Software With Free Tiers. Here is Where Each One Breaks.

Aditya Vishwakarma — Fri, 13 Nov 2026 12:58:32 +0000

The 20-minute audit that comes before any cancellation

Open your card statement and list every recurring software charge. Then mark each one with a single word:

Weekly — you opened it in the last seven days.
Monthly — you opened it this month, but not this week.
Never — you are paying for the idea of being the kind of business that uses it.

Cancel the third group today. There is no migration involved, and it is usually the largest single saving on the statement. For the second group, replace the tool with a free tier and keep the login for a month before you cancel. Only the first group has earned a subscription.

Most solo operators and freelancers find two or three tools in that first group. Not twenty. The rest of the stack is either unused or used so rarely that a free tier is not a compromise — it is the correct size.

Six replacements, and the exact place each one bites

Every free tier has a ceiling. The point is to know where it is before you hit it at the worst possible moment.

1. Client and deal tracking: HubSpot Sales Starter ($50/mo) → HubSpot Free CRM

The free tier gives you unlimited contacts, a deal pipeline, and 1,000 marketing contacts. What it does not give you is automation workflows or custom reporting — you get the pipeline, not the robot. If your follow-ups depend on automation firing, you will feel that within a week. If they depend on you remembering to follow up, the pipeline is enough.

First step: Import your spreadsheet as CSV, build one pipeline with four stages, and do not add custom properties for a week.

2. Email list: Mailchimp Standard ($20/mo) → MailerLite Free

1,000 subscribers and 12,000 emails a month, with automation on a limited set of triggers. The ceiling is real and it arrives faster than people expect. Export your list monthly as a backup so a pricing change can never hold your audience hostage.

First step: Import subscribers, then rebuild exactly one email — the welcome message. A half-built automation sequence is worse than a single good welcome email.

3. Invoicing: FreshBooks Lite ($19/mo) → Zoho Invoice Free

Unlimited invoices and clients for a single business, with no cut beyond your payment processor. Payment collection inside the invoice varies by country, so check that before you promise a client a pay-now link.

First step: Recreate your three most-used templates and set a 7-day and 14-day reminder. Most late payments are a reminder problem, not a client problem.

4. Scheduling: Calendly Standard ($12/mo) → Google Calendar appointment schedules

Free with a Google account, unlimited booking pages, buffers, and date ranges. You lose round-robin assignment and some integrations. For a solo operator, neither matters — there is nobody to round-robin to.

First step: One 30-minute page, a 15-minute buffer, and the link in your email signature. That single link removes an entire email thread from every new client conversation.

5. Screen recording: Loom Business ($15/mo) → OBS Studio

Unlimited recording length, no watermark, no monthly fee. What you give up is the instant share link: you record, then upload the file to Drive and send the link. That extra step is the entire cost of the free version.

First step: Set one scene, one microphone source, and one output folder. Record a 90-second test before you record anything a client will see.

6. Automations: Zapier Professional ($30/mo) → Make Free

1,000 operations a month and unlimited scenarios. Operations burn surprisingly fast on polling triggers, which check for new data on a schedule whether or not anything changed. Use webhooks where the tool supports them and the same budget lasts far longer.

First step: Build one automation only — a new form response creating a row and emailing you. Resist building the second one until the first has run for a week.

The two places you should keep paying

Whatever touches money directly: your payment processor, your bank, and your accountant. A free tool that mishandles a payment costs more than every subscription in this article combined. Optimise the tools around the money, not the money itself.
Anything that holds client data you could not reconstruct. Free tiers change. Export monthly, and a pricing change becomes an inconvenience instead of an emergency.

Need a 1-Page SaaS Bill Reduction Audit? ($1.00)

If you don't have the time to audit all your software subscriptions yourself, send me your software list. For $1.00 on Ko-fi, I will produce a tailored 1-Page SaaS Bill Reduction Audit within 2 hours:

Exact free-tier replacements matched to your workflow
Specific ceiling warnings where each tier will break
Step-by-step migration sequence to save $50–$200/month immediately
⚡ Request a $1.00 SaaS Bill Audit on Ko-fi
📦 Download the Full 24-Tool Stack Guide + CSV Tracker ($1.00)
💻 Browse the Open-Source Directory on GitHub
☕ Direct $1 PayPal Tip

Originally published at The Zero-Dollar Dev.

Your 8GB GPU Isn't the Problem: 4 Ollama Settings That Stop the Silent CPU Fallback

Aditya Vishwakarma — Fri, 13 Nov 2026 12:36:50 +0000

The 2 tokens/sec trap that tricks everyone into upgrading hardware

You pull a 7B or 8B model (Qwen 2.5 Coder, Llama 3.3, or DeepSeek-R1 Distill) on a machine with an 8GB RTX GPU or an M-series Mac. For the first two prompts, it screams along at 35 tokens per second. Then, on prompt three, it slows to a crawl: 2.1 tokens per second, CPU fan pinned at maximum, laptop burning hot.

Most developers assume: "My 8GB GPU is too small. I need a $1,500 RTX 4090 or a Mac Studio with 64GB unified memory."

It is almost never the GPU. In 9 out of 10 cases, your model was running 100% in VRAM until the conversation history passed a specific token boundary. At that exact moment, the unquantized KV-cache exceeded remaining VRAM by 150 megabytes, forcing Ollama to silently split layers and dump execution onto the system CPU.

The Memory Math: Where your VRAM actually goes

Total memory consumption in Ollama or llama.cpp is not just the model weights. It follows this exact formula:

Total VRAM Required = Model Weights + KV Cache Size + CUDA Context Overhead

For an 8-billion parameter model quantized at Q4_K_M:

Model weights: ~4.7 GB
CUDA runtime overhead: ~450 MB
Available VRAM left on an 8GB card: ~2.8 GB

By default, Ollama initializes the KV cache in 16-bit unquantized float (FP16). At an 8,192 token context window, FP16 KV-cache requires 2.6 GB. Add a 12,000 token system prompt or a couple of web search results, and your cache demands 3.8 GB. 4.7 GB + 3.8 GB = 8.5 GB. VRAM overflow. Instant fallback to 2 tok/s.

The 4 Settings That Fix It Today

1. Enable Quantized KV Cache (q8_0)

By default, Ollama caches key-value pairs in FP16. Setting the cache to 8-bit cuts the KV cache footprint by 50% with zero perceptible degradation in coding or reasoning benchmarks.

export OLLAMA_KV_CACHE_TYPE=q8_0

On an 8k context window, this single environment variable shrinks your memory footprint from 2.6 GB down to 1.3 GB, leaving ample headroom on 8GB cards.

2. Enable Flash Attention

Flash Attention calculates attention with linear memory complexity rather than quadratic memory spikes during long prompt evaluations. It keeps memory usage flat as conversations grow.

export OLLAMA_FLASH_ATTENTION=1

3. Cap the Context Window in Your Modelfile

Many modern models declare a 32k or 128k maximum context window in their GGUF metadata. If your tool doesn't specify a context limit, Ollama may preallocate or scale context beyond your physical memory limits.

Create a custom Modelfile and fix the context window to what your VRAM can mathematically guarantee:

FROM qwen2.5-coder:7b-instruct-q4_K_M
PARAMETER num_ctx 4096
PARAMETER temperature 0.2
PARAMETER repeat_penalty 1.15

For 95% of programming and reasoning tasks, 4,096 tokens is more than enough for a full function, error log, or code review pass.

4. Force Single-Thread Concurrency

Ollama defaults to handling multiple requests in parallel if requested by client applications. When two requests hit at once, Ollama clones the context space, instantly doubling VRAM usage.

export OLLAMA_NUM_PARALLEL=1

This guarantees that Ollama never attempts to allocate a second context buffer while running local coding agents or background tools.

How to Start Ollama With These Flags

Linux / macOS (terminal):

OLLAMA_FLASH_ATTENTION=1 OLLAMA_KV_CACHE_TYPE=q8_0 OLLAMA_NUM_PARALLEL=1 ollama serve

Windows (PowerShell):

$env:OLLAMA_FLASH_ATTENTION="1"
$env:OLLAMA_KV_CACHE_TYPE="q8_0"
$env:OLLAMA_NUM_PARALLEL="1"
ollama serve

Need a Tailored Modelfile For Your Exact Machine? ($1)

If you don't want to calculate the layer math yourself, I provide a Custom Local AI Hardware Tuning service for $1.00 on Ko-fi. Send me your CPU, GPU model/VRAM, and RAM, and within 2 hours I will generate a custom Modelfile + exact startup configuration tailored to keep your model 100% on GPU at maximum speed.

⚡ Request a $1 Custom Hardware Modelfile on Ko-fi
📦 Get the $1.00 Zero-Dollar Business Stack Download
💻 Browse the Open-Source Directory on GitHub
☕ Direct $1 PayPal Tip

Originally published at The Zero-Dollar Dev.

I Replaced $465/mo of Business Software With Free Tiers. Here is Where Each One Breaks.

Aditya Vishwakarma — Fri, 13 Nov 2026 11:23:33 +0000

The 20-minute audit that comes before any cancellation

Open your card statement and list every recurring software charge. Then mark each one with a single word:

Weekly — you opened it in the last seven days.
Monthly — you opened it this month, but not this week.
Never — you are paying for the idea of being the kind of business that uses it.

Six replacements, and the exact place each one bites

Every free tier has a ceiling. The point is to know where it is before you hit it at the worst possible moment.

1. Client and deal tracking: HubSpot Sales Starter ($50/mo) → HubSpot Free CRM

First step: Import your spreadsheet as CSV, build one pipeline with four stages, and do not add custom properties for a week.

2. Email list: Mailchimp Standard ($20/mo) → MailerLite Free

First step: Import subscribers, then rebuild exactly one email — the welcome message. A half-built automation sequence is worse than a single good welcome email.

3. Invoicing: FreshBooks Lite ($19/mo) → Zoho Invoice Free

First step: Recreate your three most-used templates and set a 7-day and 14-day reminder. Most late payments are a reminder problem, not a client problem.

4. Scheduling: Calendly Standard ($12/mo) → Google Calendar appointment schedules

First step: One 30-minute page, a 15-minute buffer, and the link in your email signature. That single link removes an entire email thread from every new client conversation.

5. Screen recording: Loom Business ($15/mo) → OBS Studio

First step: Set one scene, one microphone source, and one output folder. Record a 90-second test before you record anything a client will see.

6. Automations: Zapier Professional ($30/mo) → Make Free

First step: Build one automation only — a new form response creating a row and emailing you. Resist building the second one until the first has run for a week.

The two places you should keep paying

Whatever touches money directly: your payment processor, your bank, and your accountant. A free tool that mishandles a payment costs more than every subscription in this article combined. Optimise the tools around the money, not the money itself.
Anything that holds client data you could not reconstruct. Free tiers change. Export monthly, and a pricing change becomes an inconvenience instead of an emergency.

What I would do this week

Pick the three largest line items on your statement. Migrate one per week, not all three at once. A migration done badly costs more than the subscription you were trying to avoid — usually in lost client data or a broken invoice during a billing cycle.

Then put the saved amount somewhere you will see it. A saving you cannot see gets absorbed into the general noise of a bank account within two months.

The complete stack

I put all 24 replacements into a single document: the paid tool, the typical list price, the free replacement, the real limit of the free tier, where it bites, and the first concrete step for each one. It comes with a spreadsheet tracker so you can mark each migration done and watch the monthly total fall.

It is $1, there are no affiliate links inside, and it is an instant download: The $0 Business Stack on Ko-fi.

If you only take one thing from this article, take the audit. Knowing which subscriptions are in the "never" column is worth more than any list of alternatives.

Your 8GB GPU isn't the problem: 4 Ollama settings that stop the silent CPU fallback

Aditya Vishwakarma — Fri, 13 Nov 2026 10:32:19 +0000

You install Ollama, pull a 7B model, run it, and get 1.8 tokens/sec. Fans barely spin. nvidia-smi shows almost no VRAM used. Nothing errors out.

That's the most common local-LLM failure mode, and it isn't a broken install. Ollama decided the model didn't fit and quietly moved layers to the CPU. It rarely tells you.

Here's how to confirm it, fix it, and measure the difference.

1. Confirm the fallback

ollama ps

The PROCESSOR column is the whole story:

NAME              ID              SIZE      PROCESSOR    UNTIL
qwen2.5-coder:7b  dae161e27b0e    5.2 GB    100% CPU     4 minutes from now

100% CPU on a machine with a working GPU means the model didn't fit in VRAM at the context size Ollama asked for. Not "the GPU is too weak" — it means the allocation failed and Ollama degraded instead of refusing.

Get a baseline before changing anything:

ollama run qwen2.5-coder:7b --verbose "write a python function to debounce a callback"

The final line prints eval rate, your real tokens/sec. Write it down.

2. The real culprit is usually the KV cache, not the weights

A 7B model at Q4_K_M is roughly 4.5–5 GB of weights. On an 8 GB card that leaves ~3 GB minus whatever the desktop compositor is using — call it 2.5 GB of usable headroom.

The KV cache is what eats it. Per token, per layer:

bytes = 2 (K and V) x layers x kv_heads x head_dim x bytes_per_element

For Llama 3.1 8B (32 layers, 8 KV heads, head_dim 128) at f16 that's 128 KiB per token. Run the multiplication:

Context	KV cache (f16)
4,096	~512 MB
8,192	~1 GB
32,768	~4 GB
131,072 (model default)	~16 GB

The model file says it supports 128k context. Ollama tries to honor that. Your 8 GB card cannot. Layers spill to CPU, and you get the 1.8 tok/s.

This is why "it works fine on my friend's 16 GB Mac" — unified memory plus a larger budget changes the answer completely.

3. The four settings that actually fix it

Set these before starting the server, not after.

Linux / macOS

export OLLAMA_FLASH_ATTENTION=1
export OLLAMA_KV_CACHE_TYPE=q8_0
export OLLAMA_NUM_PARALLEL=1
ollama serve

Windows (PowerShell, persistent)

setx OLLAMA_FLASH_ATTENTION 1
setx OLLAMA_KV_CACHE_TYPE q8_0
setx OLLAMA_NUM_PARALLEL 1

Then quit Ollama from the tray and restart it — setx only affects new processes.

What each one buys you:

OLLAMA_FLASH_ATTENTION=1 — uses a fused attention kernel with a smaller memory footprint. Free win on any modern GPU.
OLLAMA_KV_CACHE_TYPE=q8_0 — quantizes the KV cache to 8-bit, roughly halving the table above. At 4k context that's ~256 MB instead of ~512 MB. Quality impact at q8 is negligible in practice; q4 KV cache is where you start noticing degradation.
OLLAMA_NUM_PARALLEL=1 — this is the one nobody mentions. Ollama defaults to serving multiple requests concurrently, and each parallel slot gets its own KV cache. On a single-user laptop that's pure waste. Setting it to 1 can free gigabytes.
num_ctx — set it explicitly, never inherit the model's declared maximum:

ollama run qwen2.5-coder:7b
>>> /set parameter num_ctx 4096
>>> /save qwen-coder-4k

Or bake it into a Modelfile:

FROM qwen2.5-coder:7b
PARAMETER num_ctx 4096
PARAMETER num_batch 512
PARAMETER temperature 0.2

ollama create qwen-coder-4k -f Modelfile

4. Pick a model that fits the budget, not the benchmark

Rule of thumb for a 4-bit quant:

weights ~= params_in_billions x 0.6 GB
+ KV cache at your chosen context
+ ~0.8 GB for the display/compositor
< total VRAM

VRAM	Realistic ceiling	Example
4 GB	3B Q4, 4k ctx	`llama3.2:3b`
6 GB	7B Q4, 4k ctx	`qwen2.5-coder:7b` at `num_ctx 4096`
8 GB	7–8B Q4, 8k ctx	`llama3.1:8b` with q8_0 KV cache
12 GB	14B Q4, 8k ctx	`qwen2.5-coder:14b`
16 GB	14B Q5 or 32B Q3	trade quality for size deliberately

A 3B model that runs at 40 tok/s on the GPU beats a 14B model that runs at 1.5 tok/s on the CPU for almost every coding loop, because you iterate more times per minute.

5. Verify the fix

ollama ps

You want 100% GPU. If it says 83%/17% CPU, you're close — drop num_ctx by half and check again.

Then re-run the same prompt from step 1 and compare eval rate. On a mobile 8 GB GPU, going from CPU fallback to full offload typically moves 7B models from ~2 tok/s to ~25–40 tok/s. That's the difference between unusable and genuinely useful.

Why this is worth ten minutes

The alternative is $20/month forever for an autocomplete box. A local 7B coder model at 30 tok/s covers most inline-completion and small-refactor work, and it costs nothing after the hardware you already own.

If you'd rather not work through the arithmetic for your specific machine, I do a $1 version: send your CPU, GPU + VRAM, RAM, and the model you want, and you get back a Modelfile plus the exact environment flags and a KV-cache headroom calculation for your hardware.

Custom tuning for your exact laptop ($1): https://ko-fi.com/seanvishwakarma/commissions
Pre-built Modelfiles for 8 GB / 16 GB / Apple Silicon ($1): https://ko-fi.com/s/124024c020
Longer write-up with the full benchmark tables: https://thezerodollardev.blogspot.com/2026/11/the-zero-dollar-local-ai-guide-how-to.html

No subscription, no account, no telemetry.