Update app.py

app.py

@@ -49,20 +49,36 @@ model_sizes = {
     "XL (180B parameters)": {"memory_required": 360},
 }
 
-def calculate_costs(instance, hours, storage, reserved, spot, years, instances):
-    data = instances[instance]
+def calculate_aws_cost(instance, hours, storage, reserved=False, spot=False, years=1):
+    data = aws_instances[instance]
     rate = data['hourly_rate']
     if spot:
-        rate *= 0.3 if instances is aws_instances else 0.2
+        rate *= 0.3
     elif reserved:
-        factors = {1: 0.6, 3: 0.4} if instances is aws_instances else {1: 0.7, 3: 0.5}
-        rate *= factors.get(years, factors[1])
-    return rate * hours + storage * (0.10 if instances is aws_instances else 0.04)
+        factors = {1: 0.6, 3: 0.4}
+        rate *= factors.get(years, 0.6)
+    compute = rate * hours
+    storage_cost = storage * 0.10
+    return {'total_cost': compute + storage_cost, 'details': data}
+
+def calculate_gcp_cost(instance, hours, storage, reserved=False, spot=False, years=1):
+    data = gcp_instances[instance]
+    rate = data['hourly_rate']
+    if spot:
+        rate *= 0.2
+    elif reserved:
+        factors = {1: 0.7, 3: 0.5}
+        rate *= factors.get(years, 0.7)
+    compute = rate * hours
+    storage_cost = storage * 0.04
+    return {'total_cost': compute + storage_cost, 'details': data}
 
 def calculate_api_cost(provider, model, in_tokens, out_tokens, calls):
     m = api_pricing[provider][model]
-    cost = in_tokens * m['input_per_1M'] + out_tokens * m['output_per_1M']
-    return cost + (calls * 0.0001 if provider == 'TogetherAI' else 0)
+    input_cost = in_tokens * m['input_per_1M']
+    output_cost = out_tokens * m['output_per_1M']
+    call_cost = calls * 0.0001 if provider == 'TogetherAI' else 0
+    return {'total_cost': input_cost + output_cost + call_cost, 'details': m}
 
 def filter_compatible(instances, min_mem):
     res = {}
@@ -70,9 +86,11 @@ def filter_compatible(instances, min_mem):
         mem_str = data['gpu_memory']
         if 'x' in mem_str and not mem_str.startswith(('1x','2x','4x','8x')):
             val = int(mem_str.replace('GB',''))
-        else:
+        elif 'x' in mem_str:
             parts = mem_str.split('x')
-            val = int(parts[0]) * int(parts[1].replace('GB','')) if len(parts)>1 else int(parts[0].replace('GB',''))
+            val = int(parts[0]) * int(parts[1].replace('GB',''))
+        else:
+            val = int(mem_str.replace('GB',''))
         if val >= min_mem:
             res[name] = data
     return res
@@ -90,67 +108,89 @@ def generate_cost_comparison(
     gcp_comp = filter_compatible(gcp_instances, min_mem)
 
     results = []
+
+    # AWS table
+    aws_html = '<h3>AWS Instances</h3>'
+    aws_html += '<table width="100%"><tr><th>Instance</th><th>vCPUs</th><th>Memory</th><th>GPU</th><th>Monthly Cost ($)</th></tr>'
     if aws_comp:
-        best_aws = min(aws_comp, key=lambda x: calculate_costs(x, compute_hours, storage_gb, reserved_instances, spot_instances, years, aws_instances))
-        cost_aws = calculate_costs(best_aws, compute_hours, storage_gb, reserved_instances, spot_instances, years, aws_instances)
-        results.append({'provider': f'AWS ({best_aws})', 'cost': cost_aws, 'type': 'Cloud'})
+        for inst in aws_comp:
+            res = calculate_aws_cost(inst, compute_hours, storage_gb, reserved_instances, spot_instances, years)
+            aws_html += f'<tr><td>{inst}</td><td>{res["details"]["vcpus"]}</td><td>{res["details"]["memory"]}GB</td><td>{res["details"]["gpu"]}</td><td>${res["total_cost"]:.2f}</td></tr>'
+        # best AWS
+        best_aws = min(aws_comp, key=lambda x: calculate_aws_cost(x, compute_hours, storage_gb, reserved_instances, spot_instances, years)['total_cost'])
+        best_aws_cost = calculate_aws_cost(best_aws, compute_hours, storage_gb, reserved_instances, spot_instances, years)['total_cost']
+        results.append({'provider': f'AWS ({best_aws})', 'cost': best_aws_cost, 'type': 'Cloud'})
+    else:
+        aws_html += '<tr><td colspan="5">No compatible instances</td></tr>'
+    aws_html += '</table>'
+
+    # GCP table
+    gcp_html = '<h3>GCP Instances</h3>'
+    gcp_html += '<table width="100%"><tr><th>Instance</th><th>vCPUs</th><th>Memory</th><th>GPU</th><th>Monthly Cost ($)</th></tr>'
     if gcp_comp:
-        best_gcp = min(gcp_comp, key=lambda x: calculate_costs(x, compute_hours, storage_gb, reserved_instances, spot_instances, years, gcp_instances))
-        cost_gcp = calculate_costs(best_gcp, compute_hours, storage_gb, reserved_instances, spot_instances, years, gcp_instances)
-        results.append({'provider': f'GCP ({best_gcp})', 'cost': cost_gcp, 'type': 'Cloud'})
-
-    api_opts = {(prov, mdl): calculate_api_cost(prov, mdl, in_tokens, out_tokens, api_calls)
-                for prov in api_pricing for mdl in api_pricing[prov]}
-    best_api = min(api_opts, key=api_opts.get)
-    results.append({'provider': f'{best_api[0]} ({best_api[1]})', 'cost': api_opts[best_api], 'type': 'API'})
-
+        for inst in gcp_comp:
+            res = calculate_gcp_cost(inst, compute_hours, storage_gb, reserved_instances, spot_instances, years)
+            gcp_html += f'<tr><td>{inst}</td><td>{res["details"]["vcpus"]}</td><td>{res["details"]["memory"]}GB</td><td>{res["details"]["gpu" ]}</td><td>${res["total_cost"]:.2f}</td></tr>'
+        best_gcp = min(gcp_comp, key=lambda x: calculate_gcp_cost(x, compute_hours, storage_gb, reserved_instances, spot_instances, years)['total_cost'])
+        best_gcp_cost = calculate_gcp_cost(best_gcp, compute_hours, storage_gb, reserved_instances, spot_instances, years)['total_cost']
+        results.append({'provider': f'GCP ({best_gcp})', 'cost': best_gcp_cost, 'type': 'Cloud'})
+    else:
+        gcp_html += '<tr><td colspan="5">No compatible instances</td></tr>'
+    gcp_html += '</table>'
+
+    # API table
+    api_html = '<h3>API Options</h3>'
+    api_html += '<table width="100%"><tr><th>Provider</th><th>Model</th><th>Input Cost</th><th>Output Cost</th><th>Total Cost ($)</th><th>Context</th></tr>'
+    api_costs = {}
+    for prov in api_pricing:
+        for mdl in api_pricing[prov]:
+            res = calculate_api_cost(prov, mdl, in_tokens, out_tokens, api_calls)
+            details = api_pricing[prov][mdl]
+            api_html += f'<tr><td>{prov}</td><td>{mdl}</td><td>${in_tokens * details["input_per_1M"]:.2f}</td><td>${out_tokens * details["output_per_1M"]:.2f}</td><td>${res["total_cost"]:.2f}</td><td>{details["token_context"]:,}</td></tr>'
+            api_costs[(prov, mdl)] = res['total_cost']
+    api_html += '</table>'
+    best_api = min(api_costs, key=api_costs.get)
+    results.append({'provider': f'{best_api[0]} ({best_api[1]})', 'cost': api_costs[best_api], 'type': 'API'})
+
+    # Recommendation and Breakeven omitted for brevity
+
+    # Chart
     df = pd.DataFrame(results)
-    aws_label = df[df['type']=='Cloud']['provider'].iloc[0]
-    gcp_label = df[df['type']=='Cloud']['provider'].iloc[1] if len(df[df['type']=='Cloud'])>1 else aws_label
-    api_label = df[df['type']=='API']['provider'].iloc[0]
-
-    fig = px.bar(df, x='provider', y='cost', color='provider', color_discrete_map={
-        aws_label: '#FF9900',
-        gcp_label: '#4285F4',
-        api_label: '#D62828'
-    })
+    colors = {r['provider']: c for r,c in zip(results, ['#FF9900','#4285F4','#D62828'])}
+    fig = px.bar(df, x='provider', y='cost', color='provider', color_discrete_map=colors)
     fig.update_yaxes(tickprefix='$')
     fig.update_layout(showlegend=False, height=500)
 
-    html = '<div></div>'  # your tables here if needed
+    html = f"""
+    <div style='padding:20px;font-family:Arial;'>
+      {aws_html}
+      {gcp_html}
+      {api_html}
+    </div>
+    """
     return html, fig
 
-def app_function(
-    compute_hours, tokens_per_month, input_ratio, api_calls,
-    model_size, storage_gb, reserved_instances, spot_instances, multi_year_commitment
-):
-    return generate_cost_comparison(
-        compute_hours, tokens_per_month, input_ratio, api_calls,
-        model_size, storage_gb, reserved_instances, spot_instances, multi_year_commitment
-    )
-
-if __name__ == "__main__":
-    with gr.Blocks(title="Cloud Cost Estimator", theme=gr.themes.Soft(primary_hue="indigo")) as demo:
-        gr.HTML('<h1 style="text-align:center;">Cloud Cost Estimator</h1>')
-        with gr.Row():
-            with gr.Column(scale=1):
-                compute_hours = gr.Slider(label="Compute Hours per Month", minimum=1, maximum=730, value=100)
-                tokens_per_month = gr.Slider(label="Tokens per Month (M)", minimum=1, maximum=1000, value=10)
-                input_ratio = gr.Slider(label="Input Ratio (%)", minimum=10, maximum=90, value=30)
-                api_calls = gr.Slider(label="API Calls per Month", minimum=100, maximum=1000000, value=10000, step=100)
-                model_size = gr.Dropdown(label="Model Size", choices=list(model_sizes.keys()), value="Medium (13B parameters)")
-                storage_gb = gr.Slider(label="Storage (GB)", minimum=10, maximum=1000, value=100)
-                reserved_instances = gr.Checkbox(label="Reserved Instances", value=False)
-                spot_instances = gr.Checkbox(label="Spot Instances", value=False)
-                multi_year_commitment = gr.Radio(label="Commitment Period (years)", choices=["1","3"], value="1")
-                submit = gr.Button("Calculate Costs")
-            with gr.Column(scale=2):
-                out_html = gr.HTML()
-                out_plot = gr.Plot()
-        submit.click(
-            app_function,
-            inputs=[compute_hours, tokens_per_month, input_ratio, api_calls,
-                    model_size, storage_gb, reserved_instances, spot_instances, multi_year_commitment],
-            outputs=[out_html, out_plot]
-        )
-    demo.launch()
+# UI setup
+with gr.Blocks(title="Cloud Cost Estimator", theme=gr.themes.Soft(primary_hue="indigo")) as demo:
+    gr.HTML('<h1 style="text-align:center;">Cloud Cost Estimator</h1>')
+    with gr.Row():
+        with gr.Column(scale=1):
+            compute_hours = gr.Slider(label="Compute Hours per Month", minimum=1, maximum=730, value=100)
+            tokens_per_month = gr.Slider(label="Tokens per Month (M)", minimum=1, maximum=1000, value=10)
+            input_ratio = gr.Slider(label="Input Ratio (%)", minimum=10, maximum=90, value=30)
+            api_calls = gr.Slider(label="API Calls per Month", minimum=100, maximum=1000000, value=10000, step=100)
+            model_size = gr.Dropdown(label="Model Size", choices=list(model_sizes.keys()), value="Medium (13B parameters)")
+            storage_gb = gr.Slider(label="Storage (GB)", minimum=10, maximum=1000, value=100)
+            reserved_instances = gr.Checkbox(label="Reserved Instances", value=False)
+            spot_instances = gr.Checkbox(label="Spot Instances", value=False)
+            multi_year_commitment = gr.Radio(label="Commitment Period (years)", choices=["1","3"], value="1")
+            submit = gr.Button("Calculate Costs")
+        with gr.Column(scale=2):
+            out_html = gr.HTML()
+            out_plot = gr.Plot()
+    submit.click(generate_cost_comparison, 
+                 inputs=[compute_hours, tokens_per_month, input_ratio, api_calls,
+                         model_size, storage_gb, reserved_instances, spot_instances, multi_year_commitment],
+                 outputs=[out_html, out_plot])
+
+demo.launch()